S.Y.M.C.

A
(TWO YEARS PATTERN)
SEMESTER - III (CBCS)

DISTRIBUTED SYSTEM AND

CLOUD COMPUTING

SUBJECT CODE : MCA32

 © UNIVERSITY OF MUMBAI
Prof. Suhas Pednekar
Vice Chancellor
University of Mumbai, Mumbai.
Prof. Ravindra D. Kulkarni Prof. Prakash Mahanwar
Pro Vice-Chancellor, Director
University of Mumbai. IDOL, University of Mumbai.

Programe Co-ordinator : Mr. Mandar L. Bhanushe

Head, Faculty of Science and Technology,
IDOL, University of Mumbai – 400098.
Course Co-ordinator : Ms. Reshma Kurkute
Assistant Professor B.Sc.IT, IDOL,
IDOL, University of Mumbai- 400098.
Course Writers : Mrs. Anjali Gaikwad Lakhamade
Bsc. IT Co-ordinator,
JES college of comm, sci and IT Jogeshwari (E) Mumbai 400060.
: Mrs Vandana Maurya
Assistant Professor,
B. K. Birla College (Autonomous), Kalyan.
: Mrs Kavita Chouk
Assistant Professor,
Satish Pradhan Dnyasadhana College.
: Dr. Krishna Sudha M
Assistant Professor Sri vasavi college, Erode.
: Mrs. Shraddha Kokate
Assistant Professor,
Ghanshyamdas Saraf College of Arts and Commerce.
: Mr Sandeep Kamble
Assistant Professor,
Cosmopolitan's Valia College.
June 2022, Print I
Published by
Director
Institute of Distance and Open Learning, University of Mumbai,Vidyanagari, Mumbai - 400 098.

DTP COMPOSED AND PRINTED BY

Mumbai University Press,
Vidyanagari, Santacruz (E), Mumbai - 400098.
 CONTENTS
Chapter No. Title Page No.

Unit I
1. Introduction To Distributed Computing Concepts 1
2. Introduction To Distributed Computing Concepts 15

Unit II
3. Clock Synchronization 33
4. Election Algorithms 47
Unit III
5. Distributed Shared Memory 59

Unit IV
6. Distributed System Management 73
7. Distributed System Management 97
8. Distributed System Management 118

Unit V
9. Introduction To Cloud Computing 138
Unit VI
10. Cloud Computing 152
11. Cloud Platforms 167
12. Cloud Issues And Challenges 180

*****
 SYLLABUS

Course Code Course Name

MCA32 Distributed System and Cloud Computing

Module Detailed Contents Hrs

1 Module: Introduction to Distributed 09
Computing Concepts:
Basic concepts of distributed systems, distributed
computing models, issues in designing distributed
systems
Inter Process Communication
Fundamental concepts related to inter process
communication including message passing
mechanism, Concepts of group communication
Remote Communication
Remote Procedural Call (RPC), Remote Method
Invocation (RMI)
 Self Learning Topics: Case study on Java RMI
2 Module: Clock synchronization: 04
Introduction of clock synchronization, Global
state, Mutual Exclusion Algorithms, Election
algorithms.
 Self Learning Topics: Synchronization in
Wireless Networks
3 Module: Distributed Shared Memory: 05
Fundamental concepts of DSM, types of DSM,
various hardware DSM systems, Consistency
models, issues in designing and implementing
DSM systems.
 Self Learning Topics: MemNet Architecture
4 Module: Distributed System Management: 06
Resource ManagementScheduling Algorithms,
Task Assignment, Load balancing approach, Load
sharing approach
Process Management
Process Migration Mechanism, Thread models
Distributed File System
Concepts of a Distributed File System (DFS), file
models
 Self Learning Topics: Case Study of anyone
distributed system
5 Module: Introduction to Cloud Computing: 06
Cloud Computing history and evolution, benefits
of cloud computing.
Cloud Computing Architecture
Cloud Architecture model, Types of Clouds:
Public Private & Hybrid Clouds, Cloud based
services: Platform as a service (PaaS), Software
as a service (SaaS), Infrastructure as a service
(IaaS)
 Self Learning Topics: Cluster computing, Grid
computing, Fog computing
6 Module: Classification of Cloud 10
Implementations:
Amazon Web Services, Microsoft Azure &
Google Cloud-- Compute Services, Storage
Services, Network Services, Database services,
Additional Services.
Google AppEngine (GAE), Aneka,
Comparativestudy of various Cloud Computing
Platforms.
Cloud Issues and Challenges
Cloud computing issues and challenges like
Security, Elasticity, Resource
 management and scheduling, QoS (Quality of
Service) and Resource Allocation, Identity and
Access Management

*****
 UNIT I

1
INTRODUCTION TO DISTRIBUTED
COMPUTING CONCEPTS
Unit Structure
1.0 Objective
1.1 Introduction
1.2 Types of distributed system
1.2.1 Client Server distributed system
1.2.2 Peer to Peer distributed system
1.3 Distributed system Overview
1.3.1 Advantages of distributed system
1.3.2 Disadvantages of distributed system
1.3.3 Challenges of distributed system
1.4 Designing issues of distributed system
1.5 Distributed system Architecture
1.6 Categories of distributed system
1.7 Distinguish between token base and non-token base algorithm
1.8 Summary
1.9 Unit End Exercise

1.0 OBJECTIVE
This chapter will able you to understand the following concept:
 Bezier curve and surface
 properties of Bezier curve
 Design techniques using Bezier curve
 Cubic Bezier curve
 Bezier surface

1.1 INTRODUCTION
1.1.1 What Is a Distributed System?
The definition of distributed system is consisting of many components
together, it can also have multiple geographic boundaries and it can also
communicate and coordinate with many components for message passing
among the actor outside this system.

1
Distributed System and
Cloud Computing

Now we will talk about the Decentralized system in which the distributed
systems are not having specific components to take the decision but every
component own their part of decision, none of them have complete
information. Hence, the resulted decision is depending upon some sort of
consensus between all components.
Distributed system is also called as parallel system as it is very close to
parallel computing. In both the terms it is been refer to scaling-up the
computational capability, but they achieve this in different way. In
parallel computing, we use multiple processors on a single machine to
perform multiple tasks simultaneously, possibly with shared memory.
Whereas in distributed computing, multiple autonomous machines with no
shared memory and communicating with message passing is used for
message passing.

1.1.2 Distributed system concepts:

1. Availability:
This term is defining the percentage of time the service is operational is
called as high availability. This is one of the most important feature of
right software.
100% availability is the feature given by best developer, achieving this
feature is the dream of every developer, it can be very challenging and
expensive.
Distributed software systems can be made up of machines with a lower
level of availability. To develop an application with 99.99% availability
you can use machines/nodes that have the four nines availability.

2. Consistency:
In this type of feature the same information can be share all nodes
simultaneously and, all nodes see and return the same information. Hence
they should work in synchronization to get all nodes to exchange
messages and work.
There some minor problem can be there like some difficulties while
passing message through the network between the nodes. Other example
of problem is in passing some message the delivery of message may fail
22
during communication or may it get lost or some nodes may be Introduction to Distributed
unavailable at some point. Computing Concepts

We can conclude that, if the function having weaker level of consistency,

the system cannot run faster – but at the same time the higher chances that
it won’t return the latest dataset.

3. Idempotency:
Idempotency means in specific request is executed when the actual event
execution will occur only one time regardless the number of times. As
long as the level of idempotency, develop by manager is try to avoid bad
consequences then it can have dropped connections, request errors, and
more.
For example, after shopping if the customer tries to make a payment but
nothing happens, he/she tries for many times, when the system is
idempotent, the payment will be charged only one time, while not using
idempotent systems one cannot give guarantee the lack of double charges
and users returning their money back.

4. Data durability:
The term durability means that once data is added to the data storage it is
one of the key concerns of distributed systems, it works even if some
system’s nodes are offline or have their data corrupted.
Level of durability is depending on different distributed databases used.
Some of them are support data durability at the machine/node level, where
as in some cases it maintains the cluster level, and in few cases it doesn’t
offer this functionality out of the box.
While developing high-scalable applications data durability takes an
important role in which it is able to process millions of events per day.
In organization world sometime the companies or the owners do not allow
the data loss as it is very crucial data. In some special cases when it deals
with critical operations and transactions. Hence the developers aim should
be providing a high level of data durability and strong connection data.
Nowadays, most distributed data storage services, e.g.
Cassandra, MongoDB, and Dynamodb, offer durability support at
different levels and can be all configured to ensure data durability at the
cluster level.

5. Message Persistence:
In message passing the many of the time it happens that while processing
a message the nodes through which a message passed goes offline or
sometime it may occur failure, then there is a risk of message loss or some
part of message loss. Message persistence assures that the message is
saved and will be processed after the issue is solved.

3
Distributed System and When we talk about quality application message persistence is one of the
Cloud Computing most important characteristics.
When we need to protect the system from losses, we can take an example
of a messaging app of Uber where billions of users are there with millions
of payments per day, it seems very difficult and requires proven
technologies and developers’ expertise.
The solution to this challenge can be a creation of a messaging system that
delivers a message at least one time and it should implement at least once.
In speaking of distributed systems, messaging is generally ensured by
some distributed messaging service like RabbitMQ or Kafka, supporting
various levels of reliability in delivering messages and allowing to build
successful app architectures.

1.2 TYPES OF DISTRIBUTED SYSTEMS

The arrangement of nodes in the distributed system is like client/server
systems or peer to peer systems.

1.2.1 Client/Server Distributed Systems:

In this type of relation, the client requests a resource to server and the
server provides that resource back to the client. Multiple clients can be
handled by the server at the same time while a client is in contact with
only one server. The computer network is must be there to communicate
with client and server in distributed system.

1.2.2 Peer to Peer Distributed Systems:

In this type of system nodes are having equal participants in data sharing
through the computer network. Every task is equally divided between all
the nodes. To complete the task nodes are interacting with each other as
required as share resources. With using the network this can be done.

1.3 DISTRIBUTED SYSTEMS OVERVIEW

1.3.1 Advantages of a Distributed System:
While distributed systems are definitely more complex to design and
build, it pays off the benefits they bring along.
● Scalability is one of the most important constraint of hardware
limitation.
● In distributed system the multiple components are used to message
passing with multiple machines hence the data can be replicate on
multiple nodes hence the Reliability is more important factor of the
distributed system as it should pass the message with reduced
capacity.

44
● The large problem is divided into smaller part so that the task can be Introduction to Distributed
perform at multiple machines at the same time is the typical Computing Concepts
applications of distributed computing. Hence, this will result into
increase in performance of many complex workloads, like matrix
multiplications.
● Transfer of data from one node to another node can be easily done as
all the nodes are connected to each other in distributed system.
● Addition of nodes at any point of time can be easily done in
distributed system i.e. it can be scaled as required.
● In the entire distributed system if one node fails the other remaining
nodes can still communicate with each other. The entire system will
not break down.
● Printers and scanner can be shared with multiple nodes rather than
being restricted to just one.

1.3.2 Disadvantages of Distributed Systems:

● The nodes and connection used in distributed system need to secured
hence it’s not provide adequate security.
● While transferring data from one node to another node messages may
have lost due to network failure.
● Single user system is much easier to handle as compare to distributed
system as the database connection is quite complicated and difficult.
● If all nodes want to send data at same time the load is on the
distributed system hence the Overloading may occur in the network.

1.3.3 Challenges in a Distributed System:

While using benefits of distributed system there are some challenges
which are discussed here.
● Consistency vs. Availability: Because of partition tolerance a
distributed system can provide either consistence or availability. It
provides by definition partition tolerance, as constrained by the CAP
theorem.
● Data Distribution: data distribution is the main challenge in network
as it uses multiple nodes to data transfer hence it need the complex
algorithm to solve the issue of distribution with effective partition.
● Coordination: the data replication can be there in distributed system
as it goes on multiple nodes hence the workload in a distributed
system is the main problem in which it is very tricky to find the
solution of the issue. To resolve the issue, it need a complex protocol
for participated nodes which can take a decision.

5
Distributed System and Often in enterprise applications, under a transaction we require multiple
Cloud Computing operations to happen at same time. For example, in a single unit of data we
may need to make several updates. While this has become quite complex
when we distribute data over a cluster of nodes. Many systems do provide
transactions like semantics in a distributed environment using complex
protocols like Paxos and Raft.

1.4 DESIGNING ISSUES OF DISTRIBUTED SYSTEM

1. Heterogeneity: In the computer network, computer hardware,
operating system and implementation of different developers is
Heterogeneity used. Client-server environment is middleware is a key
component of the heterogeneous distributed system. To communicate
with end to end customer Middleware is a set of services to interact
with the application.
2. Openness: new resources can be share and made available to the
users at any time can be known as openness of the distributed system.
Fact that their key interfaces are the main characteristics of the open
system. A uniform communication mechanism can be helpful for
access to shared resources in distributed commuting. The construction
can be done through heterogeneous hardware and software.
3. Scalability: The term Scalability can be achieved through a
significant increase in the number of users and resources connected
and its efficient use.
4. Security: The main challenge of distributed system is the security of
information. Hence the problem can be resolve with three components
Confidentially, integrity and availability and with Encryption we can
protects shared resources.
5. Failure Handling: while using system there may be some case where
some faults occur in hardware and the software program, it may
produce false results or sometime it may hold the activity before they
have finished the intended computation. bust handling failure is more
complex in distributed systems because the failure is the failure can
be in some components fail while others continue to function.
6. Concurrency: the resources can be shared at the same time by client
can lead the issue of concurrency as multiple users require the same
resources at the same time and they have the access of read, write, and
update. Every resource should be safe in a concurrent environment.

7. Transparency: Transparency ensures that the distributes system

should be perceived as a single entity by the users or the application
programmers rather than the collection of autonomous systems, which
is cooperating. The user should be unaware of where the services are
located and the transferring from a local machine to a remote one
should be transparent.

66
1.5 DISTRIBUTED SYSTEM ARCHITECTURE Introduction to Distributed
Computing Concepts
The architecture of a distributed system depends on the use-case which
can be used anywhere in the management .it shows the flow of the data.
However, with some general patterns we can explore more cases.
The following diagram can represent the fact, about the core distribution
models which the architecture used

● Master-slave: In this model, there is one node who is master node of

the distributed system and it plays the role of king. Hence he is having
all the necessary information about the system and its control and
based on these it can also take the decision whenever is needed. The
other node can act as salves and they can be responsible for the task
given by the master and they should report to the master only. The
changes in any architecture can be done by master node.
● Peer-to-peer: in peer to peer distributed system there is no master
slave relationship. The master node is not present in this system. All
nodes are master nodes in peer to peer distributed system in this
model. All the nodes equally share the responsibility of the master.
Most of the distributed system can be combination of both the architecture
i.e. Peer to peer and master slaves. One can also choose one of the
architecture for models.
Data distribution is advantage of peer-to-peer model, whereas data
replication in the same architecture can be advantage of master-slave
model.

1.6 CATEGORIES OF DISTRIBUTED SYSTEM

To design a distributed system there can be several rationales. For
example, we need to perform computations like matrix multiplications at a
massive scale in machine learning models. These are impossible to
accommodate on a single machine.
Similarly, on a single machine it is impossible to store all large files and
handling huge files and processing at least highly inefficient.
7
Distributed System and So, categorization of distributed system can be depending upon the use-
Cloud Computing case, the following categories are.
● Data stores
● Messaging
● Computing
● Ledgers
● File-systems
● Applications
Traditionally, we can store the data in relational databases were the default
choice of data store for quite simple some time. However, the data can be
growth in terms of volume, velocity of data and variety, in this
database is not sufficient for the data process hence it started to fall short
of the expectation. The solution for this problem we introduced NoSQL
databases with their distributed architecture which is more useful.
Similarly, the distributed system provides the features like durability,
performance, scalability, usability over the traditional messaging systems
which could not remain insulated to the challenges of the modern scale of
data. in this area today there are several options that can provide multiple
semantics like publish-subscribe and point-to-point.
The 5 categories of various models that are used for building distributed
computing systems

1. Minicomputer Model:

● The extension of the centralized time-sharing system is called as

minicomputer model.
● Few minicomputers interconnected by a communication network in
the distributed computing system based. multiple users can
simultaneously have logged on to each minicomputer.
88
● Each minicomputer is connected to Several interactive terminals. Introduction to Distributed
With every user login they get remote access to other minicomputers Computing Concepts
while using minicomputer.
● Every user will get an access of remote resources on network which
are available on some machine other than the one on to which the user
is currently logged. maybe when one need to share resource sharing
with remote users is desired The minicomputer model is used.
● One of the best example of distributed computing system based on the
minicomputer model is the early ARPA net.

2. Workstation Model:

● A several workstations interconnected by a communication network

in distributed computing system based on the workstation model.
● Throughout an infrastructure several workstations were located in an
organization. every workstation has its well equipped with its own
disk & serves as a single-user computer.
● In such type of environment, at any one time out of total workstations
some of the workstations are idle which results in the waste of large
amounts of CPU time.
● Therefore, the idea of the workstation model is to interconnect all
these workstations by a high-speed LAN so that idle workstations
may be used to process jobs of users who are logged onto other
workstations & do not have sufficient processing power at their own
workstations to get their jobs processed efficiently.
● Example: Sprite system & Xerox PARC.

9
Distributed System and 3. Workstation–Server Model:
Cloud Computing

● A network of personal workstations having its own hard disk and

local files can be together called as server workstation model.

● Diskful workstation can be referred as a workstation with its own

local disk & a workstation without a local disk can be referred as a
diskless workstation. in network environments these workstations
have become more popular than diskful workstations.

● A few minicomputers & several workstations interconnected by a

communication in network are together come under distributed
computing system based on the workstation-server model.

● In this model, a user logs onto a workstation with the help of his or
her home workstation. Normal algorithmic or numeric activities
required by the user's processes are performed at the user's home
workstation, but server process the requests coming from special
servers. Out of which some are sent to a server which provided by
user's requested activity & returns the result of request processing to
the user's workstation.

● Therefore, in this model, the user's processes need not migrated to the
server machines for getting the work done by those machines.

● Example: The V-System.

10
10
4. Processor–Pool Model: Introduction to Distributed
Computing Concepts

● Some time there are some cases in which the user does not need any
computing power but once in a while the user may need a very large
amount of computing power for a short time this computation can
consider under the processor-pool model. It works on the observation
of the utilization of resources.
● As we know that the workstation-server model uses a processor is
allocated to each user for the task, but in processor-pool model the
processors are used for the task to pooled together and the resources
are shared by the users as and when needed.
● a large number of microcomputers & minicomputers attached to the
network in the pool of processors.
● In this model every workstation in the pool has its own memory to
load the data & run a system program or an application program of
the distributed computing system.
● No home machine is present & the user does not log onto any
machine in this model.
● Better utilization of processing power & greater flexibility is the
highlighting advantage of the model.
● Example: Amoeba & the Cambridge Distributed Computing System.

5. Hybrid Model:
● The workstation-server model has a large number of computer users
only performing simple interactive tasks &-executing small programs.
● The processor-pool model is more attractive & suitable for the users
or the group of users who need to do massive computation for job
performing.
11
Distributed System and ● The feature of Workstation-server & processor-pool models can
Cloud Computing combine together is called as hybrid model which can be used to build
a distributed system.
● The allocation of processor can be done dynamically for computations
that are too large or require several computers for execution.
● This model assures that the interactive jobs can be processed in local
workstation of the user account in the hybrid mode.
A distributed system is a system in which components are situated in
distinct places, these distinct places refer to networked computers which
can easily communicate and coordinate their tasks by just exchanging
messages to each other. These components can communicate with each
other to conquer one common goal as a task.
There are many algorithms are used to achieve Distributed Computing and
these are broadly divided into 2 categories: Token-Based Algorithms and
Non-Token Based Algorithms.

1.7 DISTINGUISH BETWEEN TOKEN BASED AND

NON-TOKEN BASED ALGORITHMS IN DISTRIBUTED
SYSTEM
Sr. No. Token Based Algorithms Non-Token Based Algorithms
1. In the distributed computing Where as in this algorithm the
system, a unique token is token is not present and even it is
shared among all the sites in not sharing any token for
Token-based algorithm. accessing the data.
2. When the token is processes Here, two or more successive
the site is allowed to enter rounds of messages are
the into the Computer exchanged between sites to
System. determine which site is to enter
the Computer System next.
3. The request sequence order The use of timestamp in
can be used to allocate the processing the request for the
token in token based Computer Systems and to resolve
algorithm for request for the the conflict for the simultaneous
Computer Systems and to requests for the System in non-
resolve the conflict for the token based algorithm.
simultaneous requests for
the System.
4. The token-based algorithm As compared to the Token-based
produces less message Algorithm Non-Token based
traffic as compared to Non- Algorithm produces more
Token based Algorithm. message traffic.
5. They are free from deadlock They are not free from the
(i.e. here there are no two or deadlock problem as they are
more processes are in the based on timestamp only.
queue in order to wait for
12
12 messages that will actually
 can’t come) because of the Introduction to Distributed
existence of unique token in Computing Concepts
the distributed system.
6. Here, it is follow the Here there is no such type of
sequence of request execution order hence.
executed as the order they
are made in.
7. Token-based algorithms are Non-Token based algorithms are
more scalable as they can less scalable than the Token-
free your server from having based algorithms because here
to store session state and server is not free from its tasks.
also they contain all the
necessary information
which they need for
authentication.
8. Here the access control is Here the access control is not so
quite Fine-grained because fine as there is no token which
here inside the token roles, can specify roles, permission, and
permissions and resources resources for the user.
can be easily specifying for
the user.
9. Token-based algorithms Non-Token based algorithms
make authentication quite can’t make authentication easy.
easy.
10. Examples of Token-Based Examples of Non-Token Based
Algorithms are: Algorithms are:
(i) Singhal’s Heuristic (i) Lamport’s Algorithm
Algorithm (ii) Ricart-Agarwala Algorithm
(ii) Raymonds Tree Based (iii) Maekawa’s Algorithm
Algorithm
(iii) Suzuki - Kasami
Algorithm

1.8 SUMMARY
The definition of distributed system is consisting of many components
together, it can also have multiple geographic boundaries and it can also
communicate and coordinate with many components for message passing
among the actor outside this system. Distributed software systems can be
made up of machines with a lower level of availability. To develop an
application with 99.99% availability you can use machines/nodes that
have the four nines availability. The architecture of a distributed system
depends on the use-case which can be used anywhere in the management
.it shows the flow of the data. However, with some general patterns we
can explore more cases. A distributed system is a system in which
components are situated in distinct places, these distinct places refer to
networked computers which can easily communicate and coordinate their

13
Distributed System and tasks by just exchanging messages to each other. These components can
Cloud Computing communicate with each other to conquer one common goal as a task.

1.9 UNIT AND EXCERCISE

1. What is distributed system?
2. List and explain types of distributed system.
3. Explain advantages of distributed system
4. Explain challenges of distributed system
5. Explain issues while designing of distributed system
6. Explain categories of distributed system
7. Write distinguishes between token base and non-token base
algorithm.
8. Write a note on
1. Hybrid model
2. Processor pool model
3. Workstation server model

*****

14
14
 2
INTRODUCTION TO DISTRIBUTED
COMPUTING CONCEPTS
Unit Structure
2.0 Objective
2.1 Introduction
2.2 Modes of Interprocess Communication
2.2.1 Shared memory
2.2.2 Message Passing
2.2.3 Synchronization in interprocess communication
2.3 Approaches to Interprocess communication
2.4 Group Communication in distributed system
2.5 RPC in Distributed system
2.5.1 Characteristics of RPC
2.5.2 Features of RPC
2.6 Types of RPC
2.7 Architecture of RPC
2.8 Advantages & Disadvantages of RPC
2.9 Remote Method Invocation
2.10 Summary
2.11 Unit End Exercise

2.0 OBJECTIVE
This chapter will able you to understand the following concept:
● What is IPC (Inter-Process Communication) and its need
● Different modes of IPC
● Different approaches of IPC
● Group communication overview with its types
● RPC and its working
● Remote Method Invocation

2.1 INTRODUCTION

What Is IPC?
To exchange the data, the cooperating processes, need to communicate
with each this transaction of process can be called as Inter-process
communication. It is the mechanism of communicating among processes.

15
Distributed System and To share data or resources the process can be used this process refers inter
Cloud Computing - process communication or interprocess communication (IPC) and it
is specially used in operating system in computer science. This
mechanism of an operating system provides to allow the processes to
manage shared data. Typically, applications of an IPC can be client server
relationship in which client ask for resources from server and server will
reply on the client request.
In designing of microkernels and nano-kernels IPC is very important
factor, which reduce the number of functionalities provided by the kernel.
Those functionalities are then obtained by communicating with servers via
IPC, leading to a large increase in communication when compared to a
regular monolithic kernel. Variable analytic framework structures are used
in IPC. The IPC model relies on multi vector protocols which can ensure
the compatibility between them. An IPC mechanism is either synchronous
or asynchronous.

Need for IPC:

● Information Sharing: the resources sharing is the main component
as many users could demand for same information at the same time.
Thus, if there would be easiest path in resource sharing it will
available all the time.
● Computation Speedup: many of the task could perform at same time
hence the big task needs to split into several sub task for its fast
execution. This also requires related processes to exchange
information related to the task
● Modularity: Most of the time applications are built in a modular
fashion and divided into separate processes. For instance, the Google
Chrome web browser spawns a separate process for each new tab.

2.2 MODES OF INTER-PROCESS COMMUNICATION

Shared memory and message passing are the two modes through which
processes can communicate with each other. The process is divided into
several sections.

2.2.1. Shared Memory:

To established a shared memory region, require communicating process
which can run through the shared memory model. Every process has its
own memory region as well as they have its own address space to
communicate with each other, the other process which want to
communicate with them, they need to attach their address space to this
shared memory segment

16
16
 Introduction to Distributed
Computing Concepts

The above diagram explains the shared memory model of IPC. Message
sharing or information exchanging done between Process A and process B
with the help of a shared memory segment through the shared memory
region.
By default, the operating system prevents processes from accessing other
process memory. The shared memory model requires processes to agree to
remove this restriction. Besides, as shared memory is established based on
the agreement between processes, the processes are also responsible to
ensure synchronization so that both processes are not writing to the same
location at the same time.

2.2.2 Message Passing:

The shared memory model is very useful model for transferring the data,
but it is not giving the suitable outcome for some processes. some process
we can’t achieve through this for example if the data is at various system
and the process needs to exchange the data through different computer
systems in distributed computing they don’t have straightforward way to
communicate with each other in a shared memory region.
The message passing mechanism provides an alternative means processes
for communication. In this mode, processes interact with each other
through messages with assistance from the underlying operating system:

17
Distributed System and
Cloud Computing

In the above diagram two processes A, and B are communicating with

each other through message passing. Process A sends a message M to the
operating system (kernel). This message is then read by process B.
In order to successfully exchange messages, there needs to be a
communication link between the processes. There are several techniques
through which these communication links are established. Following
points discussed the mechanisms:
● Direct Communication: In this type the direct sender and receiver is
including in the process, every process has explicit the recipient or the
sender. For example, if process A needs to send a message to process
B, it can use the primitive
● Indirect Communication: In this type the messages can be
transferred indirectly with the help of, mailbox. A container that holds
the messages is called mailbox. For example, if X is a mailbox, then
process A can send a message to mailbox using the
primitive
● Synchronization: in this type the additional option of
synchronization is there which can have extension to direct and
indirect communication. There is block to send and receive message
18
18
 which can be based on the need of a process can choose to block Introduction to Distributed
while sending or receiving messages Besides, it can communicate Computing Concepts
without any blocking is called asynchronous.
● Buffering: The temporary queue for exchanged messages is called as
buffering. These queues can be of zero, bounded, and unbounded
capacity
We can explain the things with help of producer-consumer problem of
IPC. where a process is producer or we can see application is also
producers who can store data. Whereas the consumer can consume data
which ultimately an application / process by which the data is produced.
To communicate with each other the producer and consumer processes
mechanism is used.
An operating system provides mechanism of Interprocess communication
that allows processes to communicate with each other. In this
communication one process can let know to the another process that some
event is occurred or the transferring of data from one process to another.

2.2.3 Synchronization in Interprocess Communication:

Synchronization is a most important part of interprocess communication.
It can be assured by either the interprocess control mechanism or handled
by the communicating processes. Following are some of the methods to
provide synchronization −

● Semaphore:
A multiple process demands common resources at same time which is
sometime called as a semaphore with a variable that controls the access of
the resources. The Min categories of semaphores are binary semaphores
and counting semaphores.

● Mutual Exclusion:
There are some cases where the only one thread enter into critical section
that situation is called as Mutual. This technique is useful for preventing
race condition and it also can help in synchronization.

● Barrier:
As name implies barrier will not allow any individual processes to proceed
until all the processes reach it. Many parallel languages and collective
routines impose barriers.

19
Distributed System and ● Spinlock:
Cloud Computing
This is one type of lock. The checking process of lock availability is called
as spinlock where the processes trying to acquire this lock wait in a loop.
This is known as busy waiting because the process is not doing any useful
operation even though it is active.

2.3 APPROACHES TO INTER PROCESS

COMMUNICATION
● Pipe:
A unidirectional channel of data transfer is called as pipe. In this one pipe
is used for one-way data channel if both the way data transfer then we
have to use two pipes for two processes. standard input and output
methods can be used in this. Pipes are used in all POSIX systems as well
as Windows operating systems.

● Socket:
The end point for sending or receiving data in a network is called as
socket. This is true for data sent between processes on the same computer
or data sent between different computers on the same network. Most of the
operating systems use sockets for interprocess communication.

● File:
A set of data record is called as file it is stored on a disk or acquired on
demand by a file server. As per the requirement multiple processes can
access a file. Files can be used for data storage in all operating systems.

● Signal:
In a limited way of interprocess communication the signals are used. The
messages are sent from one process to another. Normally, signals are not
used to transfer data but are used for remote commands between
processes.

● Shared Memory:
Multiple process at the same time shares the resources with the help of
Shared memory. Due to this all process can communicate with one another
in easiest way All POSIX systems, as well as Windows operating systems
use shared memory.

● Message Queue:
Without being connected to each other multiple processes can read and
write data to the message queue. Until their recipient retrieves, messages
are stored in the queue. In interprocess communication Message queues
are quite useful and are used by most operating systems.

20
20
 Introduction to Distributed
Computing Concepts

2.4 GROUP COMMUNICATION OVERVIEW

Group Communication:
In group communication all the servers are connected together with each
other hence the message is sent to a group and then this message is
delivered to all server of the group.
There are so many members of group who can join or leave the group at
any time. This can be done through multicast communication. There are
three communication modes:
1. Unicast: Process to process communication
2. Broadcast: all the process can communicate with each other
Object Group: A set of
3. Multicast: only a member of group or object of a class can objects grouped together to
collectively respond to a set
communicate with each other of operations or
invocations.
Object Group: Same set of invocations concurrently applied on a set of All objects in the group are
object which is together called as an Object group. treated as a single entity
from the perspective of the
Client Group: operations on a single, local object, which acts as a proxy invoker.
for the group is called as Client objects. Invocations (methods,
commands, or operations)
The proxy uses a group communication system to send the invocations to applied to the object group
are performed concurrently
the members of the object group. or simultaneously on all
members of the group.
Various type of data (files, code, source file) can be exchanged over a
network and this is between two processes in a distributed system.
Sometime group communication concept is existing as there could be a
situation where one source process tries to communicate with multiple
processes at once. Abstraction of process or a group is a collection of
interconnected processes. In abstraction it is hiding the message passing
wo look the procedure in normal procedure. Different host works together
21
Distributed System and while communicating in Group communication and they also perform
Cloud Computing operations in a synchronized manner, so that the improvement of
performance of the system can take place.

Types of group communication:

Closed and open groups:
only a group member can send the message in group and multicast the
message in group. In the closed group a process can deliver message to
itself.

Overlapping and non-overlapping groups:

In overlapping groups, entities may be members of multiple groups, and
non-overlapping groups imply that membership does not overlap to at
most one.
In distributed computing,
every process is Types of Group Communication in a Distributed System:
connected with the host
process, and if the host Broadcast Communication:
process wants to
communicate with the In distributed computing every process is connects with the host process
other processes at the and if the host process wants to communicate with the other process at
same time, it is called same time it is called as broadcast communication. When one common
broadcast communication.
information is sent over a trusted network Broadcast communication is
used as the information is delivered to each and every process in most
efficient manner possible. It does not require any extra processing thing
the communication is very fast in comparison to other modes of
communication. It is not suitable for large number of processes and cannot
treat a specific process individually.
When one common information is sent over a trusted network, broadcast communication is used, as the
22
22 information is delivered to each and every process in the most efficient manner possible. It does not
require any extra processing, making the communication very fast in comparison to other modes of
communication. However, it is not suitable for a large number of processes, and it cannot treat a specific
process individually.
 Introduction to Distributed
Computing Concepts

A broadcast Communication: P1 process communicating with every

process in the system

Multicast Communication:
A group of process needs to controlled by a specific host process in a
distributed system at the same time is called as multicast communication.
The implementation is done in finding a way to address problem of a high
workload on host system and redundant information from process in
system. It mostly decreases the time taken for message handling.

A multicast Communication: P1 process communicating with only a

group of the process in the system

Unicast Communication:
When a single process wants to communicate with the host process in a
distributed system at the same time is called as Unicast communication.
As the name imply it deals with single process. The same information can
be pass to multiple processes. The specific process will be treating in best
way hence this is the best for two processes for communication. It will
lead to overheads as it has to find exact process and then exchange
information/data.
23
Distributed System and
Cloud Computing

Synchronous and asynchronous systems:

Group communication can be possible in both the environments. It is
reliable and ordered in multicast message processing.
In group communication, it guarantees the copy of same message should
deliver to the all members of a group.

Reliable multicast operation is based on two things:

Integrity: delivering the message correctly only once
Validity: assuring that a message sent will be delivered.
The following are the types of message ordering:
FIFO ordering: first process sends the message to first server and so on it
will be delivered in this order for all the processes in the group.
Causal ordering: If a message happens before another message in the
distributed system this so-called causal relationship will be preserved in
the delivery of the associated messages at all processes.
Total ordering: during the entire process, if a message is delivered before
another message at one process, then the same order will be preserved for
all processes is called as total ordering process.

Following are the four main task of Group membership management

Providing an interface for group membership changes: the changes
related to membership is done at this stage. The operations like addition of
process, or destroy a process from a group or crating a process can have
done through the membership service.
Failure detection: In case of crash and unreachability the service
monitors the group members. Suspected or Unsuspected processes can be
detecting by detector. If the process is unreachable or crashed the process
is excluded from membership.

24
24
Notifying members of group membership changes: if the process is Introduction to Distributed
added to the group or excluded from the group then the service will Computing Concepts
indicate that to the other group members.
Performing group address expansion: When a process multicast a
message, it supplies the group identifier rather than a list of processes in
the group.

2.5 WHAT IS RPC?

Remote Procedure Call (RPC) is an interprocess communication
technique. The extended form of RPC is Remote Procedure Call. The use
of RPC is in used for client-server applications. RPC mechanisms are used
when a computer program causes a procedure or subroutine to execute in a
different address space, which is coded as a normal procedure call without
the programmer specifically coding the details for the remote interaction.
RMI (Remote Method Invocation) is a way that a programmer, using
the Java programming language and development environment, can
write object-oriented programming in which objects on different
computers can interact in a distributed network. RMI is the Java version of
what is generally known as a remote procedure call (RPC), but with the
ability to pass one or more objects along with the request. The object can
include information that will change the service that is performed in the
remote computer. Sun Microsystems, the inventors of Java, calls this
"moving behaviour.
This procedure call also manages low-level transport protocol, such as
User Datagram Protocol, Transmission Control Protocol/Internet Protocol
etc. It is used for carrying the message data between programs.

2.5.1 Characteristics of RPC:

Here are the essential characteristics of RPC:
● The called procedure is in another process, which is likely to reside in
another machine. 25
Distributed System and ● The processes do not share address space.
Cloud Computing
● Parameters are passed only by values.
● RPC executes within the environment of the server process.
● It doesn’t offer access to the calling procedure’s environment.

2.5.2 Features of RPC:

Here are the important features of RPC:
● Simple call syntax
● Offers known semantics
● Provide a well-defined interface
● It can communicate between processes on the same or different
machines

2.6 TYPES OF RPC

There are three types of RPC:
1. Callback RPC
2. Broadcast RPC
3. Batch-mode RPC

1. Callback RPC:
P2P paradigm between participating processes can enables RPC. With this
client and server both can be work as service.

Functions of Callback RPC:

● Interactive application problems can be remotely processed
● It offers server with clients handle system
● Callback will allow the client process wait
● Will help in managing callback deadlocks
● It allows a peer-to-Peer paradigm among participating processes.

2. Broadcast RPC:
The client’s request is called as Broadcast RPC; the processing request
method can be processed by all servers on broadcast network.

Functions of Broadcast RPC:

● It allows you to specify that the client’s request message has to be
26
26 broadcasted.
● You can declare broadcast ports. Introduction to Distributed
Computing Concepts
● Physical network can be reducing with the broadcast RPC

3. Batch-mode RPC:
Batch-mode RPC helps to queue, on the client-side, separate RPC
requests, in a transmission buffer and then send them on a network in one
batch to the server.

Functions of Batch-mode RPC:

● When request is sent over the network will minimizes overhead
involved them over the network in one batch to the server.
● Only lower call rates applications are efficient for the type of RPC
protocol.
● It needs a reliable transmission protocol.

2.7 RPC ARCHITECTURE

RPC architecture has mainly five components of the program:
1. Client
2. Client Stub
3. RPC Runtime
4. Server Stub
5. Server

RPC Architecture 27
Distributed System and How RPC Works?:
Cloud Computing
Step 1) On the run time client execute, client stub, and one instance of
RPC client machine.
Step 2) A client starts a client stub process by passing parameters.
Client’s own address space stores within the client stub. It
acknowledges the server stub by local RPC Runtime.
Step 3) RPC called by the user by accessing regular Local Procedural
Call. RPC Runtime manages broadcast the messages between
the network across client and server. It also does
acknowledgment, routing, performs the job of retransmission,
and encryption.
Step 4) After getting over the server procedure, it gets back to the server
stub, in which the message values are returned. In the transport
layer the layer gets back the message by server stub.
Step 5) During the step, the client sends message result to the transport
layer, which returns back a message to the client stub.
Step 6) In this stage, the return parameters called by the client stub, in
the resulting packet, and the execution process returns to the
caller.

2.8 ADVANTAGES & DISADVANTAGES OF RPC

Advantages of RPC:
● In high-level languages RPC method helps in transmission of message
between clients and servers using the conventional procedure calls.
● on the local procedure call RPC method is modeled, but the called
procedure is most likely to be executed in a different process and
usually a different computer.
● Process and thread-oriented models can be used in RPC.
● RPC provides the feature of abstraction. The internal message passing
mechanism hidden from the user in RPC.
● The effort needs to re-write and re-develop the code is minimum.
● To improve performance, it commits many of the protocol layers.
● In a distributed environment RPC allows the usage of the
applications.

28
28
 Introduction to Distributed
Computing Concepts
Disadvantages of RPC:
● In Remote Procedure the parameter is call by values and pointer
which is not allowed in RPC.
● The time required for Remote procedure call is significantly lower
than that for a local procedure.
● The probability of failure occurs is high as it involves a
communication system, another machine, and another process.
● There are many different ways of RPC implementation, hence it is not
standard.
● Flexibility is not offered in RPC for hardware architecture as It is
mostly interaction-based.
● In remote procedure call, the cost of the process is increased.

2.9 RMI (REMOTE METHOD INVOCATION)

Remote Method Invocation:
The extended form of RMI is Remote Method Invocation. In this it
allows an object residing in one system (JVM) to gain access /invoke an
object running on another JVM.
In distributed application RMI is used. It is responsible for remote
communication between Java programs. It is provided in the
package java.rmi.

2.9.1 Goals of RMI:

● To minimize the complexity of the application.
● To preserve type safety.
● Distributed garbage collection.
● Minimize the difference between working with local and remote
objects.

2.9.2 Architecture of an RMI Application:

Server program and client program is the main feature of an RMI
application, one has to write two individual programs for each.
● In the server program it is remote object which is created and
reference of that object which can give access to the client.
● Whereas on the sever, the client program requests the remote objects
and tries to invoke its methods.
29
 Distributed System and The following diagram shows the architecture of an RMI application.
Cloud Computing

Stub:The stub is a client-side

proxy for the remote object. It
appears to the client as if it is
the actual remote object, but it
is responsible for forwarding
method calls from the client to ● Transport Layer: This layer is responsible for the connection
the actual remote object on the between the client and the server. It also responsible to manage the
server.The stub handles
communication tasks like existing connection and also sets up new connections.
marshalling parameters,
sending the request over the ● Stub: A representation (proxy) of the remote object at client is called
network, and receiving the as Stub. We will find it in the client system; it acts as a gateway for
response.The client never the client program.
directly interacts with the
remote object; instead, it calls ● Skeleton: on the server side this object is residing.
methods on the stub.
stub communicates with this skeleton to pass request to the remote
Skeleton (in older RMI object.
versions, no longer used in
Java RMI as of JDK 5):The ● RRL (Remote Reference Layer): in this layer it manages the
skeleton is a server-side references made by the client to the remote object.
component that receives the
method call from the stub, Working of an RMI Application:
unmarshalls the parameters,
and invokes the appropriate
● When the remote object gets call from the client object, it is first
method on the actual remote
object.In modern versions of received by the stub and after that it eventually passes this request to
RMI (after JDK 5), skeletons the RRL.
are no longer required, as the
server-side functionality is ● After getting request from the remote object to the client-side RRL, it
handled directly by the RMI invokes a method called invoke () of the object remoteRef. RRL gets
runtime system.
request form the server side.
● After passing the request on the server side it passes the request to the
Skeleton (proxy on the server) which finally invokes the required
object on the server.
● The result is passed all the way back to the client.

2.9.3 Marshalling and Unmarshalling:

Whenever a client invokes a method that accepts parameters on a remote
object, the parameters are bundled into a message before being sent over
the network. These parameters may be of primitive type or objects. In case
30
30
of primitive type, the parameters are put together and a header is attached Introduction to Distributed
to it. In case the parameters are objects, then they are serialized. This Computing Concepts
process is known as marshalling.
At the server side, the packed parameters are unbundled and then the
required method is invoked. This process is known as unmarshalling.

RMI Registry:
RMI registry is a namespace on which all server objects are placed. Each
time the server creates an object, it registers this object with the
RMIregistry (using bind() or reBind() methods). These are registered
using a unique name known as bind name.
To invoke a remote object, the client needs a reference of that object. At
that time, the client fetches the object from the registry using its bind name
(using lookup() method).

The following illustration explains the entire process:

2.10 SUMMARY
To exchange the data, the cooperating processes, need to communicate
with each this transaction of process can be called as Inter-process
communication. it is the mechanism of communicating among processes.
To established a shared memory region, require communicating process
which can run through the shared memory model. The message passing
mechanism provides an alternative means processes for communication.
In this mode, processes interact with each other through messages with
assistance from the underlying operating system. In group communication
all the servers are connected together with each other hence the message is
31
Distributed System and sent to a group and then this message is delivered to all server of the
Cloud Computing group. Group communication can be possible in both the environments. It
is reliable and ordered in multicast message processing. The extended
form of RMI is Remote Method Invocation. In this it allows an object
residing in one system (JVM) to gain access /invoke an object running on
another JVM. In case of primitive type, the parameters are put together
and a header is attached to it. In case the parameters are objects, then they
are serialized. This process is known as marshalling. At the server side,
the packed parameters are unbundled and then the required method is
invoked. This process is known as unmarshalling.

2.11 UNIT AND EXCERCISE

1. What is IPC? Explain Working of IPC.
2. Explain different modes of IPC
3. Explain different approaches of IPC
4. Explain RPC with its characteristics and features.
5. Explain RPC architecture with diagram.
6. List down advantages & disadvantages of RPC
7. What is RMI explain its working.
8. Explain concept of marshalling and unmarshalling
9. Write a note on RMI registry
10. Explain the concept of Group communication.

*****

32
32
 UNIT II

3
CLOCK SYNCHRONIZATION
Unit Structure
3.1 Introduction
3.2 Clock Synchronization
3.2.1 How Computer Clocks Are Implemented
3.2.2 Drifting or Clocks
3.2.3 Clock Synchronization Issues
3.2.4 Clock Synchronization Algorithms
3.3 Mutual Exclusion
3.3.1 Centralized Approach
3.3.2 Distributed Approach
3.3.3 Token-Passing Approach
3.4 Reference

3.1 INTRODUCTION
A distributed system is a collection of distinct processes that are spatially
separated and run concurrently. In systems with multiple concurrent
processes, it is economical to share the hardware or software resources
among the concurrently executing processes. In such situation, sharing
may be cooperative or competitive. Since the number of available
resources in a computing system is restricted, one process must
necessarily influence the action of other concurrently executing processes
as it competes for resources. For example, a resource such as a tape drive
that cannot be used simultaneously by multiple processes, a process
willing to use it must wait if it is in use by another process. This chapter
presents synchronization mechanisms that are suitable for distributed
systems.

3.2 CLOCK SYNCHRONIZATION

Every computer needs a timer mechanism called a computer clock to keep
track of current time, calculation of the time spent by a process in CPU
utilization, disk I/O etc. so that the corresponding user can be charged
properly. In a distributed system, an application may have processes that
concurrently run on multiple nodes of the system. For correct results, such
distributed applications require that the clocks of the nodes are
synchronized with each other. For example, for a distributed on-line
reservation system to be fair, seat booked (remaining) almost
simultaneously from two different nodes should be offered to the client
who booked first, even if the time difference between the two bookings is
33
Distributed System and very small. It may not be possible to guarantee this if the clocks of the
Cloud Computing nodes of the system are not synchronized.
In a distributed system, synchronized clocks also enable one to measure
the duration of distributed activities that start on one node and terminate
on another node, for instance, calculating the time taken to transmit a
message from one node to another at any arbitrary time. It is difficult to
get the correct result in case if the clocks of the sender and receiver nodes
are not synchronized.
The discussion above shows that it is the job of a distributed operating
system designer to devise and use suitable algorithms for properly
synchronizing the clocks of a distributed system. In this section we will
describe such algorithms. However, for a better understanding of these
algorithms, we will first discuss how computer clocks are implemented
and what are the main issues in synchronizing the clocks of a distributed
system.

3.2.1 How Computer Clocks Are Implemented:

A computer clock usually consists of three components- a quartz crystal
that oscillates at a well-defined frequency, a counter register, and a
constant register. The constant register is used to store a constant value
that is decided based on the frequency of oscillation of the quartz crystal.
The counter register is used to keep track of the oscillations of the quartz
crystal. That is, the value in the counter register is decremented by 1 for
each oscillation of the quartz crystal. When the value of the counter
register becomes zero, an interrupt is generated, and its value is
reinitialized to the value in the constant register. Each interrupt is called a
clock tick.

Figure 3.1
To make the computer clock function as an ordinary clock used by us in
our day-to day life, the following things are done:
34
34
1. The value in the constant register is chosen so that 60 clock ticks Clock Synchronization
occur in a second.
2. The computer clock is synchronized with real time (external clock).
For this, two more values are stored in the system-a fixed starting date
and time and the number of ticks. For example, in UNIX, time begins
at 0000 on January 1, 1970. At the time of initial booting, the system
asks the operator to enter the current date and time. The system
converts the entered value to the number of ticks after the fixed
starting date and time. At every clock tick, the interrupt service
routine increments the value of the number of ticks to keep the clock
running.

3.2.2 Drifting or Clocks:

A clock always runs at a constant rate because its quartz crystal oscillates
at a well-defined frequency. However, the rates at which two clocks run
are normally different from each other due to differences in the crystals.
The difference in the oscillation period between two clocks might be
extremely small, but the difference accumulated over many oscillations
leads to an observable difference in the times of the two clocks, no matter
how accurately they were initialized to the same value.
Therefore, as time passes, a computer clock drifts from the real-time clock
that was used for its initial setting. For clocks based on a quartz crystal,
the drift rate is approximately 10-6, giving a difference of 1 second every
1,000,000 seconds, or 11.6 days [Coulouris et al. 1994]. Hence there is a
need to resynchronize computer clock periodically with the real-time
clock to keep it nonfaulty. Even nonfaulty clocks do not always maintain
perfect time. A clock is considered nonfaulty if there is a bound on the
amount of drift from real time for any given finite time interval.
Suppose that when the real time is t, the time value of a clock p is C p(t). If
all clocks in the world were perfectly synchronized, we would have Cp(t)
= t for all p and all t. That is, if C denotes the time value of a clock, in the
ideal case dC/dt should be 1. Therefore, if the maximum drift rate
allowable is p, a clock is said to be nonfaulty if the following condition
holds for it:

As shown in Figure 3.2, after synchronization with a perfect clock, slow

and fast clocks drift in opposite directions from the perfect clock. This is
because for slow clocks dC/dt < 1 and for fast clocks dC/dt > 1.

35
Distributed System and
Cloud Computing

Figure 3.2 Slow, perfect, and fast clocks

A distributed system consists of several nodes, each with its own clock,
running at its own speed. Because of the nonzero drift rates of all clocks,
the set of clocks of a distributed system do not remain well synchronized
without some periodic resynchronization. This means that the nodes of a
distributed system must periodically resynchronize their local clocks to
maintain a global time base across the entire system. Recall from Figure
3.2 that slow and fast clocks drift in opposite directions from the perfect
clock.
Therefore, out of two clocks, if one is slow and one is fast, at a time
after they were synchronized, the maximum deviation between the
time value of the two clocks will be . Hence, to guarantee that no two
clocks in a set of clocks ever differ by more than , the clocks in the set
must be resynchronized periodically, with the time interval between two
synchronizations being less than or equal to . Therefore, unlike a
centralized system in which only the computer clock must be
synchronized with the real-time clock, a distributed system requires the
following types of clock synchronization:
1. Synchronization of the computer clocks with real-time (or external)
clocks:
This type of synchronization is mainly required for real-time applications.
That is, external clock synchronization allows the system to exchange
information about the timing of events with other systems and users.
An external time source UTC (Coordinated Universal Time) is often used
as a reference for synchronizing computer clocks with real time. The UTC
is an international standard and many standard bodies disseminate UTC
signals by radio, telephone, and satellite. For instance, the WWV radio
station in the United States and the Geostationary Operational
Environmental Satellites (GEOS) are two such standard bodies.
Commercial devices (called time providers) are available to receive and
interpret these signals. Computers equipped with time provider devices
can synchronize their clocks with these timing signals.
36
36
2. Mutual (or internal) synchronization of the clocks of different Clock Synchronization
nodes of the system:

This synchronization is mainly required for those applications that require

a consistent view of time across all nodes of a distributed system as well
as for the measurement of the duration of distributed activities that
terminate on a node different from the one on which they start.

Note that externally synchronized clocks are also internally synchronized.

However, the converse is not true because with the passage of time
internally synchronized clocks may drift arbitrarily far from external time.

3.2.3 Clock Synchronization Issues:

We have seen that no two clocks can be perfectly synchronized.

Therefore, in practice, two clocks are said to be synchronized at a instance
of time if the difference in time values of the two clocks is less than some
specified constant . The difference in time values of two clocks is called
clock skew.

Clock synchronization requires each node to read the other nodes clock
values. The actual mechanism used by a node to read other clocks differs
from one algorithm to another. However, regardless of the actual reading
mechanism, a node can obtain only an approximate view of its clock skew
with respect to other nodes' clocks in the system. Errors occur mainly
because of unpredictable communication delays during message passing
used to deliver a clock signal or a clock message from one node to
another. A minimum value of the unpredictable communication delays
between two nodes can be computed by counting the time needed to
prepare, transmit, and receive an empty message in the absence of
transmission errors and any other system load. However, in general, it is
rather impossible to calculate the upper bound of this value because it
depends on the amount of communication and computation going on in
parallel in the system, on the possibility that transmission errors will cause
messages to be transmitted several times, and on other random events,
such as page faults, process switches, or the establishment of new
communication routes.

An important issue in clock synchronization is that time must never run

backward because this could cause serious problems, such as the repetition
of certain operations that may be hazardous in certain cases. Notice that
during synchronization a fast clock has to be slowed down. However, if
the time of a fast clock is readjusted to the actual time all at once, it may
lead to running the time backward for that clock.

Therefore, clock synchronization algorithms are normally designed to

gradually introduce such a change in the fast running clock instead of
readjusting it to the correct time all at once. One way to do this is to make
the interrupt routine more intelligent. When an intelligent interrupt routine
is instructed by the clock synchronization algorithm to slow down its
clock, it readjusts the amount of time to be added to the clock time for
37
Distributed System and each interrupt. For example, suppose that if 8 msec is added to the clock
Cloud Computing time on each interrupt in the normal situation, when slowing down, the
interrupt routine only adds 7 msec on each interrupt until the correction
has been made. Although not necessary, for smooth readjustment, the
intelligent interrupt routine may also advance its clock forward, if it is
found to be slow, by adding 9 msec on each interrupt, instead of
readjusting it to the correct time all at once.

3.2.4 Clock Synchronization Algorithms:

Clock synchronization algorithms may be broadly classified as centralized

and distributed.

Centralized Algorithms:

In centralized clock synchronization algorithms one node has a real-time

receiver usually called the time server node, and the clock time of this
node is regarded as correct and used as the reference time. The goal of
the algorithm is to keep the clocks of all other nodes synchronized with
the clock time of the time server node. Examples of centralized algorithms
are- Berkeley Algorithm, Passive Time Server, Active Time Server etc.

Distributed Algorithms:

Distributed is the one in which there is no centralized time server present.

Instead the nodes adjust their time by using their local time and then,
taking the average of the differences of time with other nodes.

Distributed algorithms overcome the issue of centralized algorithms like

the scalability and single point failure. Examples of Distributed algorithms
are – Global Averaging Algorithm, Localized Averaging Algorithm, NTP
(Network time protocol) etc.

3.3 MUTUAL EXCLUSION

There are many resources in a system that need not be used
simultaneously by multiple processes if program operation is to be correct.
For example, a file cannot be simultaneously updated by multiple
processes. Similarly, use of tape drives or printers must be restricted to a
single process at a time. Therefore, exclusive access to such shared
resource by a process must be ensured. This exclusiveness of access is
called mutual exclusion between processes. The sections of a program that
need exclusive access to shared resources are called as critical sections.

An algorithm for implementing mutual exclusion must satisfy the

following requirements:

38
38
1. Mutual exclusion: Clock Synchronization
At any time only one process should access the shared resource. That is, a
process that has been granted the resource must release it before it can be
granted to another process.

2. No starvation:

If every process that is granted the resource eventually releases it, every
request must be eventually granted.

Three basic approaches used by different algorithms for implementing

mutual exclusion in distributed systems are described below. To simplify
our description, we assume that each process resides at a different node.

3.3.1 Centralized Approach:

In this approach, one of the processes in the system is elected as the

coordinator and coordinates the entry to the critical sections. Each process
that wishes to enter a critical section must first seek permission from the
coordinator. If no other process is currently in that critical section, the
coordinator can immediately grant permission to the requesting process.
However, if two or more processes concurrently ask for permission to
enter same critical section, the coordinator grants permission to only one
process at a time in accordance with some scheduling algorithm. After
executing a critical section, when a process exits the critical section, it
must notify the coordinator process so that the coordinator can grant
permission to another process (if any) that has also asked for permission to
enter the same critical section.

An algorithm for mutual exclusion using centralized approach is described

here with the help of an example. As shown in Figure 3.3, assume that
there is a coordinator process Pc and three other processes P1, P2 and P3 in
the system. Also assume that the requests are granted in the first-come,
first-served manner for which the coordinator maintains a request queue.
Suppose P1 wants to enter a critical section for which it sends a request
message to Pc. On receiving the request message, Pc checks to see whether
some other process is currently in that critical section. Since no other
process is in the critical section, Pc immediately sends back a reply
message granting permission to Pl. When reply arrives, P1 enters critical
section.

39
Distributed System and
Cloud Computing

Figure 3.3 Example illustration centralize approach for mutual

exclusion
Now suppose that while P1 is in the critical section and at same time P2
asks for permission to enter the same critical section by sending a request
message to Pc. Since P1 is already in the critical section, permission cannot
be granted to P2. The exact method used to deny permission varies from
one algorithm to another. For our algorithm, let us assume that the
coordinator does not return any reply and the process that made the
request remains blocked until it receives the reply from the coordinator.
Therefore, Pc does not send a reply to P2 immediately and enters its
request in the request queue.
Again, assume that while P1 is still in the critical section and at same time
P3 also sends a request message to Pc asking for permission to enter same
critical section. Obviously, P3 cannot be granted permission, so no reply is
sent immediately to P3 by Pc; and its request is queued in the request
queue.
Now suppose P1 exits the critical section and sends a release message to
Pc releasing its exclusive access to the critical section. On receiving the
release message, Pc takes the first request from the queue of deferred
requests and sends a reply message to the corresponding process, granting
it permission to enter the critical section. Therefore, in this case, Pc sends a
reply message to P2.
After receiving the reply message, P2 enters the critical section, and when
it exits the critical section, it sends a release message to Pc. Again Pc takes
the first request from the request queue (in this case request of P3 ) and
40
40 sends a reply message to the corresponding process (P3) On receiving the
reply message, P3 enters the critical section, it sends a release message to Clock Synchronization
Pc, when it exits the critical section. Now since there are no more requests,
Pc keeps waiting for the next request message.
This algorithm ensures mutual exclusion because, at a time, the
coordinator allows only one process to enter a critical section. Use of first-
come, first-served scheduling policy ensures that no starvation will occur.
This algorithm is simple to implement and requires only three messages
per critical section entry: a request, a reply, and a release.
However, it suffers from the usual drawbacks of centralized schemes. That
is, a single coordinator is subject to a single point of failure and can
become a performance bottleneck in a large system. Furthermore, for
handling failure, mechanism must be provided to detect a failure of the
coordinator, to elect a unique new coordinator, and to reconstruct its
request queue before the computation can be resumed.

3.3.2 Distributed Approach:

In the distributed approach, the decision making for mutual exclusion is
distributed across the entire system. All processes that want to enter the
same critical section cooperate with each other before reaching to a
decision on which process will enter the critical section next. The first
such algorithm was presented by Lamport [1978] based on his event-
ordering scheme. Later, Ricart and Agrawala [1981] proposed a more
efficient algorithm which also requires that there should be a total ordering
of all events in the system. As an example of a distributed algorithm for
mutual exclusion, Ricart and Agrawala's algorithm is described below. In
the following description we assume that Lamport's event ordering scheme
is used to generate a unique timestamp for each event in the system.
When a process wants to enter a critical section, it sends a request message
to all other processes. The message contains the following information:
1. The process identifier of the process.
2. The name of the critical section that the process wants to enter.
3. A unique timestamp generated by the process for the request message.
On receiving a request message, a process either immediately sends back a
reply message to the sender or defers sending a reply based on the
following rules:
1. If the receiver process is currently executing in the critical section, it
simply queues the request message and defers sending a reply.
2. If the receiver process is currently not executing in the critical section
but is waiting for its turn to enter the critical section, it compares the
timestamp in the received request message with the timestamp in its
own request message that it has sent to other processes. If the
timestamp of the received request message is lower, it means that the
sender process made a request before the receiver process to enter the
41
Distributed System and critical section. Therefore, the receiver process immediately sends
Cloud Computing back a reply message to the sender. On the other hand, if the receiver
process's own request message has a lower timestamp, the receiver
queues the received request message and defers sending a reply
message.
3. If the receiver process neither is in the critical section nor is waiting
for its turn to enter the critical section, it immediately sends back a
reply message.
A process that sends out a request message keeps waiting for reply
messages from other processes. It enters the critical section as soon as it
has received reply messages from all processes. After finishing execution
in the critical section, it sends reply messages to all processes in its queue
and deletes them from its queue.
Let us consider the example of Figure 3.4, to illustrate how the algorithm
works. There are four processes P1, P2, P3 and P4. While process P4 is in a
critical section, processes P1 and P2 want to enter the same critical section.
To get permission from other processes, processes P1 and P2 send request
messages with timestamps 6 and 4 respectively to other processes (Figure
3. 4(a)).

Figure 3.4 Example illustrating the distributed algorithm for mutual

exclusion
(a) status when processes P1 and P2 send request messages to other
42
42 processes while process P4 is already in the critical section;
(b) status while process P4 is still in critical section; Clock Synchronization
(c) status after process P4 exits critical section;
(d) status after process P2 exits critical section.
Now let us consider the situation in Figure 3.4(b). Since process P4 is
already in the critical section, it defers sending a reply message to P 1 and
P2 and enters them in its queue. Process P3 is currently not interested in the
critical section, so it sends a reply message to both P1 and P2.
Process P2 defers sending a reply message to P1 and enters P1 in its queue,
because the timestamp (4) in its own request message is less than the
timestamp (6) in P1'S request message. On the other hand, P1 immediately
replies to P2 because the timestamp (6) in its request message is found to
be greater than the timestamp (4) of P2'S request message.
Next consider the situation in Figure 3.4(c). When process P4 exits the
critical section, it sends a reply message to all processes in its queue (in
this case to processes P1 and P2) and deletes them from its queue. Now
since process P2 has received a reply message from all other processes (P1,
P3 and P4), it enters the critical section. However, process P1 continues to
wait since it has not yet received a reply message from process P2.
Finally, when process P2 exits the critical section, it sends a reply message
to P1 (Fig. 3.4(d). Since process P1 has received a reply message from all
other processes, it enters the critical section.
This algorithm guarantees mutual exclusion because a process can enter its
critical section only after getting permission from all other processes, and
in the case of a conflict, only one of the conflicting processes can get
permission from all other processes. Algorithm also ensures freedom from
starvation since entry to the critical section is scheduled according to the
timestamp ordering.
It has also been proved by Ricart and Agrawala [1981] that the algorithm
is free from deadlock. Furthermore, if there are n processes, the algorithm
requires n-1 request messages and n-1 reply messages, giving a total of
2(n-1) messages per critical section entry.
This algorithm suffers from the following drawbacks because of the
requirement that all processes must participate in a critical section entry
request by any process:
1. In a system having n processes, the algorithm is liable to n points of
failure because if process fails, the entire scheme collapses. This is
because the failed process will not reply to request messages that will
be falsely interpreted as denial of permission by the requesting
processes, causing all the requesting processes to wait indefinitely.
To solve this problem Tanenbaum [1995] proposed a small
modification to the algorithm. In the modified algorithm, instead of
remaining silent by deferring the sending of the reply message in
43
Distributed System and cases when permission cannot be granted immediately, the receiver
Cloud Computing sends a "permission denied" reply message to the requesting process
and then later sends an OK message when the permission can be
granted. Therefore, a reply message (either "permission denied" or
OK) is immediately sent to the requesting process in any case. If the
requesting process does not receive a reply from a process within a
fixed time period, it either keeps trying until the process replies or
concludes that the process has crashed. When the requesting process
receives a "permission denied" reply message from one or more of
the processes, it blocks until OK message is received from all of
them.
2. The algorithm requires that each process knows the identity of all the
processes participating in the mutual-exclusion algorithm. This
requirement makes implementation of the algorithm complex because
each process of a group needs to dynamically keep track of the
processes entering or leaving the group. That is, when a process joins
a group, it must receive the names of all the other processes in the
group, and the name of the new process must be distributed to all the
other processes in the group. Similarly, when a process leaves the
group or crashes, all members of that group must be informed so that
they can delete it from their membership list. Updating of the
membership list is little bit difficult when request and reply messages
are already being exchanged among the processes of the group.
Therefore, the algorithm is suitable only for groups whose member
processes are fixed and do not change dynamically.
3. In this algorithm, a process can enter a critical section only after
communicating with all other processes and getting permission from
them. Therefore, assuming that the network can handle only one
message at a time, the waiting time from the moment the process
makes a request to enter a critical section until it actually enters the
critical section is the time for exchanging 2(n-1) messages in a system
having n processes. This waiting time may be large if there are too
many processes in the system. Therefore, the algorithm is suitable
only for a small group of cooperating processes.

3.3.3 Token-Passing Approach:

In this approach, mutual exclusion is achieved by using a single token that
is circulated among the processes in the system. A token is a special type
of message that entitles its holder to enter a critical section. For fairness,
the processes in the system are logically organized in a ring structure, and
the token is circulated from one process to another around the ring always
in the same direction (clockwise or anticlockwise).
When a process receives the token, it checks if it wants to enter a critical
section and acts as follows:
 If it wants to enter a critical section, it keeps the token with it, enters
the critical section, and exits from the critical section after finishing
44
44 its work. It then passes the token along the ring to its neighbor
 process. Note that the process can enter only one critical section when Clock Synchronization
it receives the token. If it wants to enter another critical section, it
must wait until it gets the token again.
 If it does not want to enter a critical section, it just passes the token
along the ring to its neighbor processes. Therefore, if no process is
interested in entering a critical section, the token simply keeps on
circulating around the ring.
Mutual exclusion is guaranteed by this algorithm because at any instance
of time only one process can be in a critical section, since there is only a
single token. Furthermore, since the ring is unidirectional and a process is
permitted to enter only one critical section each time it gets the token,
starvation cannot occur.
In this algorithm the number of messages per critical section entry may
vary from 1 (when every process always wants to enter a critical section)
to an unbounded value (when no process wants to enter a critical section).
Moreover, for a total of n processes in the system, the waiting time from
the moment a process wants to enter a critical section until its actual entry
may vary from the time needed to exchange 0 to n-1 token-passing
messages. Zero token-passing messages are needed when the process
receives the token just when it wants to enter the critical section, whereas
n-1 messages are needed when the process wants to enter a critical section
just after it has passed the token to its neighbor process.
The algorithm requires the handling of the following types of failures:

1. Process failure:
A process failure in the system causes the logical ring to break. In such
situation, a new logical ring must be established to ensure the continued
circulation of the token among other processes. This requires detection of
a failed process and dynamic reconfiguration of the logical ring when a
failed process is detected or when a failed process recovers after failure.
Failed process can be easily detected by making it a rule that a process
receiving the token from its neighbor always sends an acknowledgment
message to its neighbor. With this rule, a process detects that its neighbor
has failed when it sends the token to it but does not receive the
acknowledgment message within a fixed time period.
On the other hand, dynamic reconfiguration of the logical ring can be done
by maintaining the current ring configuration with each process. When a
process detects that its neighbor has failed, it removes the failed process
from the group by skipping it and passing the token to the next alive
process in the sequence. When a process becomes alive after recovery, it
simply informs its previous neighbor in the ring so that it gets the token
during the next round of circulation.

45
Distributed System and 2. Lost token:
Cloud Computing
If the token is lost, a new token must be generated. The algorithm must
have mechanisms to detect and regenerate a lost token. One method to
solve this problem is to designate one of the processes on the ring as a
"monitor" process. The monitor process periodically circulates a "who
has the token?" message on the ring. This message rotates around the
ring from one process to another.
All processes simply pass this message to their neighbor process, except
the process that has the token when it receives this message. This process
writes its identifier in a special field of the message before passing it to its
neighbor. After one complete round, when the message returns to the
monitor process it checks the special field of the message. If there is no
entry in this field, it concludes that the token has been lost, generates a
new token, and circulates it around the ring.
There are two problems associated with this method- the monitor process
may itself fail and the "who has the token?" message may itself get
lost. Both problems may be solved by using more than one monitor
processes. Each monitor process independently checks the availability of
the token on the ring. However, when a monitor process detects that the
token is lost, it holds an election with other monitor processes to decide
which monitor process will generate and circulate a new token. An
election is required to prevent the generation of multiple tokens that may
happen when each monitor process independently detects that the token is
lost, and each one generates a new token.

3.4 REFERENCE
 Pradeep K. Sinha, Distributed Operating System: Concepts and
Design, PHI Learning.

*****

46
46
 4
ELECTION ALGORITHMS
Unit Structure
4.1 Deadlock Introduction
4.1.1 Necessary Conditions for Deadlock
4.1.2 Deadlock Modeling
4.1.3 Deadlock Prevention
4.2 Election Algorithms
4.2.1 The Bully Algorithm
4.2.2 A Ring Algorithm
4.2.3 Discussion of the Election Algorithms
4.3 Summary
4.4 Reference

4.1 DEADLOCK
In the previous section we saw that there are several resources in a system
for which the resource allocation policy must ensure exclusive access by a
process. A system consists of a finite number of units of each resource
type for example, three printers, six tape drives, four disk drives, two
CPUs, etc. When multiple concurrent processes have to compete to use
resource, the sequence of events required to use a resource by a process is
as follows:

1. Request:
The process first makes a request for the resource. If the requested
resource is not available, possibly because it is being used by another
process, the requesting process must wait until the requested resource is
allocated to it by the system.
Note that if the system has multiple units of the requested resource type,
the allocation of any unit of the type will satisfy the request. Process may
request as many units of a resource as it requires with the restriction that
the number of units requested may not exceed the total number of
available units of the resource.

2. Allocate:
The system allocates the resource to the requesting process as soon as
possible. It maintains a table in which it keeps records of each resource
whether it is free or allocated and, if it is allocated, to which process its
allocated. If the requested resource is currently allocated to another
process, the requesting process is added to a queue of processes waiting
for this resource. Once the system allocates the resource to the requesting
process, that process can exclusively use the resource by operating on it.
47
 Distributed System 3. Release:
and Cloud Computing
Once the process has finished using the allocated resource, it releases the
resource to the system. The system table records are updated at the time of
allocation and release to reflect the current status of availability of
resources.
The request and release of resources are called system calls, such as
request and release for devices, open and close for files, and allocate and
free for memory space. Notice that of the three operations, allocate is the
only operation that the system can control. The other two operations are
initiated by a process. If the total request made by multiple concurrent
processes for resources of a certain type exceeds the available amount,
some way is required to order the assignment of resources in time. Extra
care must be taken that the strategy applied cannot cause a deadlock, that
is, a situation where a set of processes are blocked because each process is
holding a resource and waiting for another resource acquired by some
other process.
It may happen that some processes that have entered the waiting state
(because the requested resources were not available at the time of request)
will never again change state, because the resources they have requested
are held by other waiting processes. This situation is called deadlock, and
the processes involved are said to be deadlocked. Hence, deadlock is the
state of permanent blocking of a set of processes each of which is waiting
for an event that only another process in the set can cause.
A deadlock situation can be explained with the help of an example given
below. Suppose that a system has two processes P1 and P2 & two tape
drives T1 and T2; resource allocation strategy is such that a requested
resource is immediately allocated to the requester if the resource is free.
Suppose that two concurrent processes P1 and P2 make requests for the
tape drives in the following order:
1. P1 requests for one tape drive and the system allocates tape drive T1 to
it.
2. P2 requests for one tape drive and the system allocates tape drive T2 to
it.
3. Now P1 requests for one more tape drive and enters a waiting state
because no tape drive is currently available.
4. Now P2 requests for one more tape drive and it also enters a waiting
state because no tape drive is currently available.
Now onwards, P1 and P2 will wait for each other indefinitely, since P1
will not release T1 until it gets T2 to carry out its designated task, that is,
not until P2 has released T2, whereas P2 will not release T2 until it gets
T1. Therefore, the two processes are in a state of deadlock. Note that the
requests made by the two processes are totally legal because each is
requesting for only two tape drives, which is the total number of tapes
48
48
drives available in the system. However, the deadlock problem occurs Election Algorithms
because the total requests of both processes exceed the total number of
units for the tape drive and the resource allocation policy is such that it
immediately allocates a resource on request if the resource is free.
In the context of deadlocks, the term "resource" applies not only to
physical objects such as tape and disk drives, printers, CPU cycles, and
memory space but also to logical objects such as a locked record in a
database, files, tables, semaphores, and monitors. However, these
resources should permit only exclusive use by a single process at a time
and should be nonpreemptable. A nonpreemptable resource is one that
cannot be taken away from a process to which it was allocated until the
process voluntarily releases it. If taken away, it has ill effects on the
computation already performed by the process. For example, a printer is a
nonpreemptable resource because taking the printer away from a process
that has started printing but has not yet completed its printing job and
giving it to another process may produce printed output that contains a
mixture of the output of the two processes. This is certainly unacceptable.

4.1.1 Necessary Conditions for Deadlock:

Coffman et al. [1971] stated that the following conditions are necessary
for a deadlock situation to occur in a system:

1. Mutual-exclusion:
If a resource is held by a process, any other process requesting for that
resource must wait until the resource has been released.

2. Hold-and-wait:
Processes can request for new resources without releasing the resources
that they are currently holding.

3. No-preemption:
A resource that has been allocated to a process becomes available for
allocation to another process only after it has been voluntarily released by
the process holding it.

4. Circular-wait:
Two or more processes must form a circular chain in which each process
is waiting for a resource that is held by another process.
A set of processes are waiting for each other in a circular fashion. For
example, let’s say there are a set of processes {P0, P1, P2, P3} such
that P0 depends on P1, P1 depends on P2, P2 depends on P3 and P3
depends on P0. This creates a circular relation between all these processes,
and they have to wait forever to be executed.
All four conditions must hold simultaneously in a system for a deadlock to
occur. If anyone of them is absent, no deadlock can occur. Notice that the
four conditions are not completely independent because the circular-wait
49
 Distributed System condition implies the hold-and-wait condition. Although these four
and Cloud Computing conditions are somewhat interrelated, it is quite useful to consider them
separately to devise methods for deadlock prevention.

Example:

Figure 4.1 Deadlock example

In the above figure, there are two processes and two resources. Process 1
holds "Resource 1" and needs "Resource 2" while Process 2 holds
"Resource 2" and requires "Resource 1". This creates a situation of
deadlock because none of the two processes can be executed. Since the
resources are non-shareable, they can only be used by one process at a
time (Mutual Exclusion). Each process is holding a resource and waiting
for the other process to release the resource it requires. None of the two
processes releases their resources before their execution and this creates a
circular wait. Therefore, all four conditions are satisfied.

4.1.2 Deadlock Modeling:

Deadlocks can be modeled using directed graphs. Before presenting a
graphical model for deadlocks, some terminology from graph theory is
needed:

1. Directed graph:
A directed graph is a pair (N, E), where N is a nonempty set of nodes and
E is a set of directed edges. A directed edge is an ordered pair (a, b),
where a and b are nodes in N.

2. Path:
A path is a sequence of nodes (a, b, c, ......i, j) of a directed graph such that
(a, b), (b, c), ... , (i, j) are directed edges. Obviously, a path contains at
least two nodes.

3. Cycle:
A cycle is a path whose first and last nodes are the same.

50
50
4. Reachable set: Election Algorithms

The reachable set of a node a is the set of all nodes b such that a path
exists from a to b.

5. Knot:
A knot. is a nonempty set K of nodes such that the reachable set of each
node in K is exactly the set K. A knot always contains one or more cycles.
An example of a directed graph is shown in Figure 4.2.
The graph has a set of nodes {a, b, c, d, e, f} and a set of directed edges
{(a, b), (b, c), (c, d), (d, e), (e, f) (f, a), (e, b)}. It has two cycles (a, b, c, d,
e, f, a) and (b, c, d, e, b). It also has a knot {a, b, c, d, e, f} that contains
the two cycles of the graph.

Figure 4.2 A directed graph

For deadlock modeling, a directed graph, called a resource allocation
graph, is used in which both the set of nodes and the set of edges are
partitioned into two types, resulting in the following graph elements:

1. Process nodes:
A process node represents a process of the system. In a resource allocation
graph, it is normally shown as a circle, with the name of the process
written inside the circle (nodes P1, P2, and P3 of Fig. 4.3)

2. Resource nodes:
A resource node represents a resource of the system. In a resource
allocation graph, it is normally shown as a rectangle with the name of the
resource written inside the rectangle. Since a resource type R, may have
more than one unit in the system, each such unit is represented as a bullet
within the rectangle. For instance, in the resource allocation graph of

51
 Distributed System Figure 4.3, there are two units of resource R1, one unit of R2, and three
and Cloud Computing units of R3.

3. Assignment edges:
An assignment edge is a directed edge from a resource node to a process
node. It signifies that the resource is currently held by the process. In
multiple units of a resource type, the tail of an assignment edge touches
one of the bullets in the rectangle to indicate that only one unit of the
resource is held by that process. Edges (R1, P1), (R1, P3), and (R2, P2)
are the three assignment edges in the resource allocation graph of Figure
4.3.

4. Request edges:
A request edge is a directed edge from a process node to a resource node.
It signifies that the process made a request for a unit of the resource type
and is currently waiting for that resource. Edges (P1, R2) and (P2, R1 ) are
the two request edges in the resource allocation graph of Figure 4.3.

Figure 4.3 Resource allocation graph

A resource allocation graph provides an overall view of the processes
holding or waiting for the various resources in the system. Therefore, the
graph changes dynamically as the processes in the system request for or
release resources or the system allocates a resource to a process. That is,
when a process Pi requests for a unit of resource type Rj , a request edge
52
52
(Pi, Rj ) is inserted in the resource allocation graph. When this request can Election Algorithms
be fulfilled, a unit of resource Rj is allocated to Pi and the request edge
(Pi, Rj ) is instantaneously transformed to an assignment edge (Rj , Pi).
Later, when Pi releases Rj , the assignment edge (Rj , Pi) is deleted from
the graph.
4.1.3 Deadlock Prevention:
We can prevent Deadlock by eliminating any of the below four conditions.
1. Mutual-exclusion
2. Hold-and-wait
3. No-preemption
4. Circular-wait

1. Eliminate Mutual Exclusion:

It is not possible to dis-satisfy the mutual exclusion because some
resources, such as the tape drive and printer, are inherently non-shareable.
Let’s take a practical example to understand this issue. Jack and Jones
share a bowl of soup. Both wants to drink the soup from the same bowl
and use a single spoon simultaneously, which is not feasible.
2. Eliminate Hold and wait:
1. Allocate all required resources to the process at the start of its
execution, this way hold and wait condition is eliminated but it will
lead to low device utilization. For example, if a process requires
printer later and we have allocated printer before the start of its
execution printer will remain blocked till it has completed its
execution.
2. The process will make a new request for resources after releasing the
current set of resources. This solution may lead to starvation.

Figure 4.4

53
 Distributed System 3. Eliminate No Preemption:
and Cloud Computing
Preempt resources from the process when resources are required by other
high priority processes.
4. Eliminate Circular Wait:
Each resource will be assigned a numerical value, and a process has to
access the resource in increasing or decreasing order. For Example, if
process P1 is allocated resource R5, now next time if P1 ask for R4 or R3
which is lesser than R5; this request will not be granted, only request for
resources greater than R5 will be granted.

Distributed system:
Distributed system is a collection of independent computers that do not
share their memory. Each processor has its own memory and they
communicate via communication networks. Communication in networks
is implemented in a process on one machine communicating with a
process on another machine.
Distributed Algorithm is an algorithm that runs on a distributed system.
Several distributed algorithms require that there be a coordinator process
in the entire system that performs some type of coordination activity
needed for the smooth running of other processes in the system. Since all
other processes in the system must interact with the coordinator, they all
must unanimously agree on who the coordinator is. Furthermore, if the
coordinator process fails due to the failure of the site on which it is
located, a new coordinator process must be elected to take up the job of
the failed coordinator.

4.2 ELECTION ALGORITHMS

Election algorithms are meant for electing a coordinator process from the
currently running processes in such a manner that at any instance of time
there is a single coordinator for all processes in the system.

Election algorithms are based on the following assumptions:

1. Each process in the system has a unique priority number.
2. Whenever an election is held, the process having the highest priority
number among the currently active processes is elected as the
coordinator.
3. On recovery, a failed process can take appropriate actions to rejoin the
set of active processes.
Therefore, whenever initiated, an election algorithm finds out which of the
currently active processes has the highest priority number and then it
informs this to all other active processes. Different election algorithms
differ in the way they do this. Two such election algorithms are described

54
54
below. In the description of both algorithms we will assume that there is Election Algorithms
only one process on each node of the distributed system.

4.2.1 The Bully Algorithm:

This algorithm was proposed by Garcia-Molina in 1982. In this algorithm
it is assumed that every process knows the priority number of every other
process in the system. The algorithm works as follows.
When any process (say Pi) sends a request message to the coordinator and
does not receive a reply within a fixed time period, it assumes that the
coordinator has failed. It then initiates an election by sending an election
message to every process with a higher priority number than itself. If Pi
does not receive any response to its election message within a fixed time
period, it assumes that it has the highest priority number among the
currently active processes.
Therefore, it takes up the job of the coordinator and sends a message (let
us call it a coordinator message) to all processes having lower priority
numbers than itself, informing that it is elected as their new coordinator.
On the other hand, if Pi receives a response for its election message, this
means that some other process having higher priority number is alive.
Therefore, Pi does not take any further action and just waits to receive the
result (a coordinator message from the new coordinator) of the election it
initiated.
When a process (say Pj ) receives an election message (obviously from a
process having a lower priority number than itself), it sends -a response
message (let us call it alive message) to the sender informing that it is
alive and will take over the election activity. Now Pj holds an election if it
is not already holding one. In this way, the election activity gradually
moves on to the process that has the highest priority number among the
currently active processes and eventually wins the election and becomes
the new coordinator.
As part of the recovery action, this method requires that a failed process
(say Pk ) must initiate an election on recovery. If the current coordinator's
priority number is higher than that of Pk, then the current coordinator will
win the election initiated by Pk and will continue to be the coordinator. On
the other hand, if Pk's priority number is higher than that of the current
coordinator, it will not receive any response for its election message. So, it
wins the election and takes over the coordinator's job from the currently
active coordinator. Therefore, the active process having the highest
priority number always wins the election.
Hence the algorithm is called the "bully" algorithm. If the process having
the highest priority number recovers after a failure, it does not initiate an
election because it knows from its list of priority numbers that all other
processes in the system have lower priority numbers than that of its own.
Therefore, on recovery, it simply sends a coordinator message to all other
processes and bullies the current coordinator into submission.

55
 Distributed System Now we will see working of this algorithm with the help of an example.
and Cloud Computing Suppose the system consists of five processes P1, P2, P3, P4 and P5 and
their priority numbers are 1, 2, 3, 4, and 5 respectively. Also assume that
at an instance of time the system is in a state in which P2 is crashed, and
P1, P3, P4 and P5 are active. Starting from this state, the functioning of the
bully algorithm with the changing system states is illustrated below.
1. P5 is the coordinator in the starting state. Suppose P5 crashes.
2. Process P3 sends a request message to P5 and does not receive a reply
within the fixed time period.
3. Process P3 assumes that P5 has crashed and it initiates an election by
sending an election message to P4 and P5. An election message is sent
only to processes with higher priority numbers.
4. When P4 receives P3's election message, it sends an alive message to
P3, informing that it is alive and will take over the election activity.
Process P5 cannot respond to P3's election message because it is down.
5. Now P4 holds an election by sending an election message to P5.
6. Process P5 does not respond to P4's election message because it is
down, and therefore, P4 wins the election and sends a coordinator
message to P1, P2 and P3 informing them that from now onwards it is
new coordinator. Obviously, this message is not received by P2
because it is currently down.
7. Now suppose P2 recovers from failure and initiates an election by
sending an election message to P3, P4 and P5. Since P2's priority
number is lower than that of P4 (current coordinator), P4 will win the
election initiated by P2 and will continue to be the coordinator.
8. Finally, suppose P5 recovers from failure. Since P5 is the process with
the highest priority number, it simply sends a coordinator message to
P1, P2, P3 and P4 and becomes the new coordinator.

4.2.2 A Ring Algorithm:

The following algorithm is based on the ring-based election algorithms
presented in [Tanenbaum 1995, Silberschatz and Galvin 1994]. In this
algorithm it is assumed that all the processes in the system are organized
in a logical ring. The ring is unidirectional means all messages related to
the election algorithm are always passed only in one direction
(clockwise/anticlockwise).
Every process in the system knows the .structure of the ring, so while
circulating a message over the ring, if the successor of the sender process
is down, the sender can skip over the successor, or the one after that, until
an active member is found. The algorithm works as follows.
When a process (say Pi) sends a request message to the current coordinator
and does not receive a reply within a fixed time period, it assumes that the
56
56
coordinator has crashed. Therefore, it initiates an election by sending an Election Algorithms
election message to its successor (to the first successor that is currently
active). This message contains the priority number of process Pi. On
receiving the election message, the successor appends its own priority
number to the message and passes it on to the next active member in the
ring. This member appends its own priority number to the message and
forwards it to its own successor.
In this manner, the election message circulates over the ring from one
active process to another and eventually returns to process Pi. Process P i
recognizes the message as its own election message by seeing that in the
list of priority numbers held within the message. The first priority number
is its own priority number.
Note that when process Pi receives its own election message, the message
contains the list of priority numbers of all processes that are currently
active. Therefore, of the available processes in this list, it elects the
process having the highest priority number as the new coordinator. It then
circulates a coordinator message over the ring to inform all the other
active processes who the new coordinator is. When the coordinator
message comes back to process Pi after completing its one round along the
ring, it is removed by process Pi. At this point all the active processes
know who the current coordinator is.
When a process (say Pj) recovers after failure, it creates an inquiry
message and sends it to its successor. The message contains the identity of
process Pj. If the successor is not the current coordinator, it simply
forwards the enquiry message to its own successor. Hence the inquiry
message moves forward along the ring until it reaches the current
coordinator. On receiving an inquiry message, the current coordinator
sends a reply to process Pj informing that it is the current coordinator.
In this algorithm two or more processes may almost simultaneously
discover that the coordinator has crashed and then each one may circulate
an election message over the ring. Although this results in a little waste of
network bandwidth, it does not cause any problem because every process
that initiated an election will receive the same list of active processes, and
all of them will choose the same process as the new coordinator.

4.2.3 Discussion of the Election Algorithms:

In the bully algorithm, in a system having total n processes, when the
process having the lowest priority number detects the coordinator's failure
and initiates an election, altogether n-2 elections are performed one after
another for the initiated one. That is, all the processes, except the active
process with the highest priority number and the coordinator process that
has just failed, perform elections by sending messages to all processes
with higher priority numbers. Hence, in the worst case, the bully algorithm
requires O(n2 ) messages. However, when the process having the priority
number just below the failed coordinator detects that the coordinator has
failed, it immediately elects itself as the coordinator and sends n-2

57
 Distributed System coordinator messages. Hence, in the best case, the bully algorithm requires
and Cloud Computing only n-2 messages.
On the other hand, in the ring algorithm, irrespective of which process
detects the failure of the coordinator and initiates an election, an election
always requires 2(n-l) messages (assuming that only the coordinator
process has failed); n-l messages are needed for one round rotation of the
election message, and another n-l messages are needed for one round
rotation of the coordinator message.
Next let us consider the complexity involved in the recovery of a process.
In the bully algorithm, a failed process must initiate an election on
recovery. Therefore, once again depending on the priority number of the
process that initiates the recovery action, the bully algorithm requires
O(n2) messages in the worst case, and n-l messages in the best case.
On the other hand, in the ring algorithm, a failed process does not initiate
an election on recovery but simply searches for the current coordinator.
Hence, the ring algorithm requires only n/2 messages on an average for
recovery action.
In conclusion, as compared to the bully algorithm, the ring algorithm is
more efficient and easier to implement.

4.3 SUMMARY
Several distributed algorithms require that there should be a coordinator
process in the entire system. Election algorithms are meant for electing a
coordinator process from among the currently running processes. We
discussed two election algorithms in this chapter i.e the bully algorithm
and the ring algorithm.

4.4 REFERENCE
 Pradeep K. Sinha, Distributed Operating System: Concepts and
Design, PHI Learning.

*****

58
58
 UNIT III

5
DISTRIBUTED SHARED MEMORY
Unit Structure
5.1 Introduction of Distributed Shared Memory
5.2 Fundamentals concept of DSM
5.2.1 Architecture of DSM
5.2.2 Advantages of DSM
5.2.3 Disadvantages of DSM
5.2.4 Types of Distributed Shared Memory (DSM)
5.2.5 Various hardware DSM systems
5.2.6 Consistency Models
5.2.7 Issues in designing and implementing DSM systems
5.3 Summery
5.4 Reference for further reading
5.5 Bibliography
5.6 Further Reading topics
5.7 Question

5.1 INTRODUCTION
DSM stands Distributed Shared Memory. It is form of memory
architecture where shared means the address space of memory shared.
DSM is a mechanism of allowing user processes to access shared data
without using inter-process communications. DSM refers to shared
memory paradigm applied to loosely coupled distributed memory systems.
Shared memory exists virtual only which is similar virtual memory so
sometimes it’s also called DSVM-Distributed Shared Virtual Memory. It
provides a virtual address space shared among processes on loosely
coupled processors.
It is basically an abstraction that integrates the local memory of different
machine into a single logical entity shared by cooperating processes.
In DSM, every implement shared node has its own memory and provides
memory read and write services and it provide consistency protocols. The
distributed shared memory (DSM) implements shared memory model in
distributed system but doesn’t have physical shared memory. All the
nodes share the virtual address space provided by the shared memory
model.

60
Distributed Shared memory: Distributed Shared Memory

1) Message Passing Paradigm:

Send (recipient, data)
Receive(data)

2) Shared Memory Paradigm-

Data=Read(address)
Write (address, data)

DSM can be achieved via software as well as hardware. DSM Provide

service memory reads and write from any node’s consistency protocols.

5.2 FUNDAMENTAL CONCEPTS OF DSM

Definition: Distributed Shared Memory is mechanism allowing end user
processes to access shared data without using inter process
communications. DSM system is to make inter process communication
transparent to end user. By using DSM variables can shared directly and
cost of communication is invisible.

. 61
 Distributed System Disturbed shared memory comparison with Message passing
and Cloud Computing
Message passing Distributed shared memory
Variables must be marshalled. Variables are shared directly
Cost of communication is visible Cost of communication is invisible
Processes are protected by having Process could cause error by altering
private address space data.
Process executes at same time. Process executing may happen with
non-overlapping lifetimes.

5.2.1 Architecture of DSM:

Each node of the system consists of one or more CPU’s and memory unit.
Nodes are connected by high-speed communication network. Simple
message passing system for nodes to exchange information. Main memory
of individual nodes is used to cache pieces of shared memory space. Main
memory of individual nodes is used to cache pieces of shared memory
space.

Memory Mapping Manger Unit:

Memory mapping manager routine maps local memory to shared virtual
memory. For mapping operation, the shared memory house is divided into
blocks. DMA uses information caching to scale back network latency.
Memory mapping manager of every node reads its native memory as
enormous cache of the shared memory house for its native process.

Communication Network Unit:

Once method access information within shared address house mapping
maps the shared memory address to the physical memory. Physical
memory on every node holds pages of shared virtual address house native
pages area in some node’s memory.

62
62
In simple language we can say nodes can communicate through managing Distributed Shared Memory
application. Refers the following diagram of distributed shared memory
communication among nodes.

5.2.2 Advantages of DSM:

1) DSM system is scalable. Scales are good with many nodes.
2) It hides the message passing.
3) DSM can handle large and complex data bases without replication or
sending them processes.
4) It provides large virtual memory space.
5) It provides portable as they use common DSM programming
interface.
6) DSM can improve performance by speeding up data access.
7) DSM provides on demand data movement means data will eliminate
the data exchange phase.
8) DSM is less expensive when compared to using a multiprocessor
system.
9) No memory access bottleneck as no single bus.

5.2.3 Disadvantages of DSM:

1) It could cause a performance penalty.
2) It should provide for protection against simultaneous access to shared
data such as lock.
3) Its performance of irregular problems could be difficult.

63
 Distributed System 5.2.4 Types of Distributed Shared Memory (DSM)
and Cloud Computing
1) On- Chip Memory
2) Bus-Based Multiprocessors
3) Ring-Based Multiprocessors
4) Switched Multiprocessors

1) On- Chip Memory:

The data is present in the CPU portion of the chips.
Address lines are connected to memory.
On-chip memory DSM is complex and expensive.
On-chip Memory chips are widely used in appliances cars and even toys.

2) Bus-Based Multiprocessors:
A bus-based multiprocessor
is a type of multiprocessor A set of parallel wires called bus acts as a connection between CPU and
system where multiple memory.
processors (CPUs) are
connected to a single shared Network traffic is reduced by using caches with each CPU.
communication bus, which is
used to access shared Some Algorithms are used to prevent two CPU trying to access same
memory or communicate with
other devices. This memory simultaneously.
use case : Small-scale multiprocessor systems
architecture is commonly where the number of processors is limited (e.g.
used in small-scale As uses single bus makes it overloaded. , 2-8 processors).
multiprocessor systems due
to its simplicity and cost-
effectiveness.

Advantages :
Simplicity
cost-Effective
Ease of programming

DisAdvantages :
Scalability issue
contention : multiple
processors
64
64 competing for the
bus can lead to delay
Latency :
 3) Ring-Based Multiprocessors: Distributed Shared Memory

A single address line is partitioned into a private area and shared area.
All nodes are connected via a token passing ring.
Shared area is divided into 32 bytes.
There is no global centralized memory present in ring-based
multiprocessors.
Communication
Mechanism: Data
travels along the ring,
passing through
intermediate
processors to reach its
destination.

Ring Interconnect: A
circular connection
structure linking all
processors in the
system.

4) Switched Multiprocessors: The concept of

Switched
Multiprocessors refers
Two approaches can be taken to attack the problem of not enough to a system
bandwidth. architecture where
multiple processors
Reduce the amount of communication Ex. Caching (CPUs) work together
to perform tasks more
Increase the communication capacity Ex. Changing topology. efficiently.

To build the system as a hierarchy build the system as multiple clusters.

However, a common
challenge in such
systems is ensuring
that there's enough
communication
bandwidth to support
data exchanges
between processors

65
 Distributed System 5.2.5 Various hardware DSM systems:
and Cloud Computing
Distributed shared memory (DSM) can be implemented in software or
hardware.
Software implementation of distributed shared memory is easy, and which
is used by software. Message passing on same cluster, so it is easy. When
we apply software distributed shared memory there is no need to change
hardware.

Software layer can be:

1) Page Based
2) Shared Variable Based
3) Object Based
Whenever we talk about hardware it relies on interconnects physically.
● For hardware distributed shared memory network interfaces and cache
coherence circuits are used.
● Special interfaces are used that supports shared memory operations.
● It gives higher performance.
● Hardware distributed shared memory is expensive.
Operating System Manges physical memory. Operating system running on
node that is physical one which access physical memory which allows us
to make connection between virtual to physical address of other nodes.
Network interface card (NIC) used to translate remote message access to
messages. NIC manages memory management, consistency, and atomics.

In this diagram,
CPU-Central Processing Unit
Mm-Memory Management
NIC-Network Interface Card

66
66
5.2.6 Consistency Models: Distributed Shared Memory

Consistency Model nothing but the degree of consistency that must be

maintained for shared memory data. Memory (states) accordingly by
access ordering and propagation or visibility of updated.
Classification of consistency model

Notations:

R-m1(X)= X was read from memory location m1

W-m1(Y)=Y was written to memory location m1

Consistency Model types:

1) Strict Consistency Model
2) Sequential Consistency Model
3) Causal Consistency Model
4) Weak Consistency Model
5) Pipelined Random Access Consistency Model (PRAM)
6) Processor Consistency Model
7) Release Consistency Model

1) Strict Consistency Model:

This type of model has strong type of memory integration with a strong
need for consistency. The amount returned for the performance learned at
the memory address remains the same as the value written for the most
transaction at that address. For writes its visible to all the process.

67
 Distributed System
and Cloud Computing

Let’s us consider the transaction example for strict consistency model

2) Sequential Consistency Mode:

The shared memory system is said to support sequential compatibility
model when all processes detect the same order. If three functions are read
(r1), write (w1), read (r2) are done in the memory field for that program
any orders
(r1,w1,r2),(r1,r2,w1),(w1,r1,r2),(w1,r2,r1),(r2,r1,w1),(r2,w1,r1) these all
processes are in same orders.

3) Causal Consistency Model:

All processes detect only those memory reference functions in their
sequence that may be related memory reference functions may be
identified by various process in different sequences. The function of the
memory reference is said to be related to other memory reference function
if one may be influenced by the other.

4) Weak Consistency Model:

If any process caused an error or any process in critical phase, there is no
restriction to show the change in the memory caused by an activity in
other process. The DSM system that supports a weak static model uses a
special alternative called synchronization variable.

5) Pipelined Random Access Consistency Model (PRAM):

It provides a simplistic semantics that is weaker than the symmetry model
described ensure that all writing process performed in one process are
recognized by all other process in sequence. All single process writing
tasks are on track.

68
68
6) Processor Consistency Model: Distributed Shared Memory

This is like Pipelined Random Access Consistency Model (PRAM).

Memory consistency means that in memory location all process agrees on
the same sequence of all writing activities performed in same memory
space. If two processes w1, w2 writing in same memory location with
sequence w1, w2 or w2, w1.

7) Release Consistency Model:

To enhance weak consistency, model this model is used. acquire used to
tell the system it is entering release used to tell the system just excited
Results are release by process of another node.
Let’s consider the transaction for release consistency model

5.2.7 Issues in designing and implementing DSM systems:

1) Granularity
2) Structure of shared memory
3) Memory Coherence and access synchronization
4) Data Location and access
5) Replacement Strategy
6) Thrashing
7) Heterogeneity

1) Granularity:
Granularity means block size of shared memory. When data sharing and
transfer across the network that time block faults occurs may be word or
page faults. A chunk of memory contains the word and issue is size of the
chunk.

69
 Distributed System
and Cloud Computing

2) Structure of shared memory:

It is layout of shared data in the memory. This structure of shared memory
depends on different application which is supported by Distributed Shared
Memory (DSM).

3) Memory Coherence and access synchronization:

Distributed Shared Memory (DSM) system allows duplicate of shared data
items, copied of shared data items available in main memories of number
of nodes. To deal with this issue the consistency of a piece of shared data
lying in the main memories of two or more nodes.

4) Data Location and access:

As shared memory required shared data, to shared data in a Distributed
Shared Memory (DSM) to locate and retrieve the data accessed by a user
process.

5) Replacement Strategy:
In this strategy the data block must be replaced by new data block. When
the local memory of the node is full, a cache misses at that node from
fetched of access of data block from remote node and replacement also.

6) Thrashing:
Data block of memory migrate between nodes on demands. Different
technique used to less thrashing.
Application Controlled Lock
Algorithm to shared date used pattern a

70
70
 page access by using Distributed Shared Memory

two variables with same page

Suppose we must access pages by using different variables we can use
same pages. consider two variables respectively and b.

6) Heterogeneity:
It is applied on operating system, computer hardware, networks on system
with implementation of developers. Heterogeneity provides environment
like client-server .it treats as middleware which is set of services which
interacts to end users. Basically, Heterogeneity works on
1) Networks: Internet protocols is used in network for communication
purpose.
2) Computer Hardware: Internal representation of different processor.
3) Programming Language: Data structured are represented for data
using different programming languages.
4) Operating System:To communicate different operating system used
to send message.
5) Implementation of different Developers: Follows different
standards for communication.

5.3 SUMMARY
DSM means Distributed Shared Memory which is not physical memory it
is virtual memory. It is a form of memory architecture where shared
means the address space of memory shared. DSM is a mechanism of
allowing user processes to access shared data without using inter-process
communications. DSM refers to the shared memory paradigm applied to
loosely coupled distributed memory systems. We learned what is
Distributed shared memory with its architecture that what is the role of
memory mapping manager and communication network unit. What is the
difference between message passing and distributed shared memory?
There are different types of distributed shared memory likes On- Chip
Memory, Bus-Based Multiprocessors, Ring-Based Multiprocessors and
Switched Multiprocessors.
As Distributed shared memory can be implemented in software as well as
in hardware. In software there are three layers Page Based, Shared
Variable Base, Object Based. In Hardware CPU (Central Processing Unit),
MU (Memory Unit) and NIC (Network Interface Card) required. We
learned different Consistency Models of distributed shared memory.
We discussed different issues of distributed shared memory.

71
 Distributed System 5.4 BIBLIOGRAPHY
and Cloud Computing
1)
https://www.cc.gatech.edu/classes/AY2010/cs4210_fall/papers/DS
M_protic.pdf
2) https://www.geeksforgeeks.org/what-is-distributed-shared-memory-
and-its-advantages/
3) https://vedveethi.co.in/eNote/DistSys/CS-702%20U-2.htm
4) https://en.wikipedia.org/wiki/Distributed_shared_memory
5) https://www.slideserve.com/dianne/distributed-computing
6) https://slideplayer.com/slide/6897641/
7) https://nptel.ac.in/courses/106102114/
8) Pradeep K. Sinha, Distributed Operating System: Concepts and
Design, PHI Learning, ISBN No. 978-81-203-1380-4
9) Andrew S. Tanenbaum, Distributed Operating Systems, Pearson
Education, ISBN No. 978-81-317-0147-8

5.6 REFERENCE FURTHER READING TOPICS

MemNet Architecture We learned in types of distributed shared memory
in this one of the types is ring based multiprocessor that is MemNet. You
can learn in details studies of Memnet with its architecture.

5.7 QUESTIONS
1) Explain Architecture of Distributed Shared Memory.
2) What is Distributed shared memory? Explain its architecture.
3) Explain various consistency model in details.
4) Explain Different types of Distributed Shared Memory.
5) What is the issue of design and implementation of distributed shared
memory?
6) Explain advantages and disadvantages of distributed shared memory.
7) Explain Distributed shared memory with its architecture?

*****

72
72
 UNIT IV

6
DISTRIBUTED SYSTEM MANAGEMENT
Unit Structure
6.1 Introduction to distributed system
6.1.1 Types of Distributed Systems
6.1.2 Advantages of Distributed Systems
6.1.3 Disadvantages of Distributed Systems
6.1.4 How does a distributed system work?
6.2 Introduction to resource management
6.2.1 Architecture of resource Management in distributed
Environment
6.3 Scheduling algorithms
6.3.1 Features of scheduling algorithms
6.3.2 Local Scheduling
6.3.3 Stride Scheduling
6.3.3.1 Extension to Stride Scheduling
6.3.4 Predictive Scheduling
6.3.5 Coscheduling
6.3.6 Gang Scheduling
6.3.7 Implicit Coscheduling
6.3.8 Dynamic Coscheduling
6.4 Task assignment Approach
6.4.1 Resource Management
6.4.2 Working of Task Assignment Approach
6.4.3 Goals of Task Assignment Algorithms
6.4.4 Need for Task Assignment in a Distributed System
6.4.5 Example of Task Assignment Approach
6.5 Load balancing approach
6.5.1 Classification of Load Balancing Algorithms
6.5.2 Issues in Load Balancing Algorithms
6.6 Load sharing approach
6.6.1 Basic idea
6.6.2 Load Estimation Policies
6.6.3 Process Transfer Policies
6.6.4 Differences between Load Balancing and Load Sharing
6.7 Summary
6.8 Reference for further reading
6.9 Model Questions

73
Distributed System and 6.1 INTRODUCTION TO DISTRIBUTED SYSTEM
Cloud Computing
A distributed system contains multiple nodes that are physically separate
but linked together using the network. All the nodes in this system
communicate with each other and handle processes in tandem. Each of
these nodes contains a small part of the distributed operating system
software.

6.1.1 Types of Distributed Systems:

The nodes in the distributed systems can be arranged in the form of
client/server systems or peer to peer systems. Details about these are as
follows −
Client/Server Systems: In client server systems, the client requests a
resource and the server provides that resource. A server may serve
multiple clients at the same time while a client is in contact with only one
server. Both the client and server usually communicate via a computer
network and so they are a part of distributed systems.
Peer to Peer Systems: The peer to peer systems contains nodes that are
equal participants in data sharing. All the tasks are equally divided
between all the nodes. The nodes interact with each other as required as
share resources. This is done with the help of a network.

6.1.2 Advantages of Distributed Systems:

● All the nodes in the distributed system are connected to each other. So
nodes can easily share data with other nodes.
● More nodes can easily be added to the distributed system i.e. it can be
scaled as required.
● Failure of one node does not lead to the failure of the entire
distributed system. Other nodes can still communicate with each
other.
● Resources like printers can be shared with multiple nodes rather than
being restricted to just one.

6.1.3 Disadvantages of Distributed Systems:

● It is difficult to provide adequate security in distributed systems
because the nodes as well as the connections need to be secured.
● Some messages and data can be lost in the network while moving
from one node to another.
● The database connected to the distributed systems is quite
complicated and difficult to handle as compared to a single user
system.
● Overloading may occur in the network if all the nodes of the
74
74 distributed system try to send data at once.
6.1.4 How does a distributed system work?: Distributed System
Management
Distributed systems have evolved over time, but today’s most common
implementations are largely designed to operate via the internet and, more
specifically, the cloud. A distributed system begins with a task, such as
rendering a video to create a finished product ready for release. The web
application, or distributed applications, managing this task — like a video
editor on a client computer — splits the job into pieces. In this simple
example, the algorithm that gives one frame of the video to each of a
dozen different computers (or nodes) to complete the rendering. Once the
frame is complete, the managing application gives the node a new frame
to work on. This process continues until the video is finished and all the
pieces are put back together. A system like this doesn’t have to stop at just
12 nodes — the job may be distributed among hundreds or even thousands
of nodes, turning a task that might have taken days for a single computer
to complete into one that is finished in a matter of minutes.
There are many models and architectures of distributed systems in use
today. Client-server systems, the most traditional and simple type of
distributed system, involve a multitude of networked computers that
interact with a central server for data storage, processing or other common
goal. Cell phone networks are an advanced type of distributed system that
share workloads among handsets, switching systems and internet-based
devices. Peer-to-peer networks, in which workloads are distributed among
hundreds or thousands of computers all running the same software, are
another example of a distributed system architecture. The most common
forms of distributed systems in the enterprise today are those that operate
over the web, handing off workloads to dozens of cloud-based virtual
server instances that are created as needed, then terminated when the task
is complete.

6.2 INTRODUCTION TO RESOURCE MANAGEMENT

Distributed systems contain a set of resources interconnected by a
network. Processes are migrated to fulfill their resource requirements. The
Resource manager is responsible to control the assignment of resources to
processes. The resources can be logical (shared file) or physical (CPU)
resource. Scheduling is the way in which processors are assigned to run
on the available resources. In a distributed computing system, the
scheduling of various modules on particular processing nodes may be
preceded by appropriate allocation of modules of the different tasks to
various processing nodes and then only the appropriate execution
characteristic can be obtained. The task allocation becomes the important
most and major activity in the task scheduling within the operating system
of a DCS.
6.2.1 Architecture of Resource Management In Distributed
Environment:
In this architecture all the processes involved in a task like resource
management play similar roles interacting co-operatively as peers without
75
Distributed System and any distinction between client and server processes or the computers they
Cloud Computing run . The aim of peer to peer architecture is to exploit the resources in a
large number of participating computers for the fulfilment of a given task.
In this architecture applications are composed of large numbers of peer
processes running on separate computers and the pattern of
communication between them depends entirely on application
requirements. A large number of data objects like files are shared. Each
object or file is replicated in several computers to further distribute the
load and to provide compensation in the event of disconnection or network
failure.
The user interface for distributed resource environment needs several
kinds of transparency for resources which are distributed on many systems
connected to the network of the distributed system. Access transparency
and location transparency are general concept for distributed resources.
The advantage of the distributed environment is the replication of the
resources in any site. Besides the two transparencies, there is semantics
transparency in which a user can access resource by semantic name rather
than by physical name for the resource.
Besides the distributed architecture of Resource Management, Scheduling
is done to make resources available to an application whenever they are
needed. Processes need to have resources assigned to them according to
priority. A resource Scheduler determines the priority of processes based
on certain criteria. Scheduling methods need to be applied to all resources
that affect the performance of an application.
There are two kinds of scheduling. Fair Scheduling and Real-Time
Scheduling. The scheduling also depends upon the types of resources. A
resource is any object that a process can request and wait for. A resource
can consist of any number of identical units and a process can request any
number of units of a resource. The types of resources can be given as:
● Reusable Resources: A reusable resource does not vanish as a result
of its use but can be used over again and again. In a system, a reusable
resource has a fixed number of units and these units can neither be
created nor destroyed.
● Consumable Resources: A consumable resource vanishes as a result
of its use. When a unit of consumable resource is allocated to a
process it as consumed and ceases to exist. There is no fixed number
of units of a consumable resource in a system since the resource units
can be created and consumed.
In a distributed resource management system, resources can be stored at
any machine and the computation can be performed at any machine. When
a machine needs to access a file stored on remote machine, the remote
machine performs the necessary file access operations and returns the data
if a read operation is performed. The two most important services present
in a distributed resource management system are the name server and
cache manager. A name server is a process that maps names specified by
76
76 clients to stored objects such as files and directories. A cache manager is a
process that implements file caching. In file caching, a copy of data stored Distributed System
at file server is brought to the client’s machine when referenced by the Management
client. Subsequent accesses to data are performed locally at the client,
there by reducing the access delays due to network latency. Cache
managers can be present at both clients and file servers. Cache managers
at the server caches files in the main memory to reduce the delays due to
disk latency.
Typcally, accessing remote resources is more expensive than accessing
local resources because of the communication delay that occur in the
network and the CPU overhead incurred to process communication
protocols. The impetus behind the development of Distributed Computing
was the availability of powerful processors at low cost and also advances
in communication technology. The availability of powerful yet cheap
processors led to the development of powerful workstations that satisfy a
single user’s needs. These powerful standalone workstations satisfy user
need by providing such things as bit-mapped displays and visual
interfaces, which traditional time-sharing systems do not support. When a
group of people work together, there is generally a need to communicate
with each other, to share data, and to share expensive resources such as
printers, disk drives etc. This requires interconnecting computers and
resources.
Technically, a completely pure peer-topeer application must implement
only peering protocols that do not recognize the concepts of server‖ and
―client‖. Such pure peer applications are rare. Most networks and
application described as peer-to-peer actually contain or rely on some non-
peer.

6.3 SCHEDULING ALGORITHMS

Scheduling is a decision-making process that is used on a regular basis in
many manufacturing and services industries. It deals with the allocation of
resources to tasks over given time periods and its goal is to optimize one
or more objectives

6.3.1 Features of Scheduling algorithms:

General purpose:
● A scheduling approach should make few assumptions about and have
few restrictions to the types of applications that can be executed.
● Interactive jobs, distributed and parallel applications, as wellas non-
interactive batch jobs, should all be supported with good performance.
● This property is a straightforward one, but to some extent difficult to
achieve.
● Because different kinds of jobs have different attributes, their
requirements to the scheduler may contradict.
● To achieve the general purpose, a tradeoff may have to be made.
77
Distributed System and Efficiency:
Cloud Computing
● It has two meanings: one is that it should improve the performance of
scheduled jobs as much as possible; the other is that the scheduling
should incur reasonably low overhead so that it would not counter
attack the benefits.

Fairness:
● Sharing resources among users raises new challenges in guaranteeing
that each user obtains his/her fair share when demand is heavy is
fairness.
● In a distributed system, this problem could be exacerbated such that
one user consumes the entire system.
● There are many mature strategies to achieve fairness on a single node.

Dynamic:
● The algorithms employed to decide where to process a task should
respond to load changes, and exploit the full extent of the resources
available.

Transparency:
● The behavior and result of a task’s execution should not be affected
by the host(s) on which it executes.
● In particular, there should be no difference between local and remote
execution.
● No user effort should be required in deciding where to execute a task
or in initiating remote execution; a user should not even be aware of
remote processing.
● Further, the applications should not be changed greatly.
● It is undesirable to have to modify the application programs in order
to execute them in the system.

Scalability:
● A scheduling algorithm should scale well as the number of nodes
increases.
● An algorithm that makes scheduling decisions by first inquiring the
workload from all the nodes and then selecting the most lightly loaded
node has poor scalability.
● This will work fine only when there are few nodes in the system.
● This is because the inquirer receives a flood of replies almost
simultaneously, and the time required to process the reply messages
78
78
 for making a node selection is too long as the number of nodes (N) Distributed System
increase. Management

● Also the network traffic quickly consumes network bandwidth.

● A simple approach is to probe only m of N nodes for selecting a node.

Fault tolerance:
● A good scheduling algorithm should not be disabled by the crash of
one or more nodes of the system.
● Also, if the nodes are partitioned into two or more groups due to link
failures, the algorithm should be capable of functioning properly for
the nodes within a group.
● Algorithms that have decentralized decision making capability and
consider only available nodes in their decision making have better
fault tolerance capability.

Quick decision making capability:

● Heuristic methods requiring less computational efforts (and hence less
time) while providing near-optimal results are preferable to
exhaustive (optimal) solution methods.

Balanced system performance and scheduling overhead:

● Algorithms that provide near-optimal system performance with a
minimum of global state information (such as CPU load) gathering
overhead are desirable.
● This is because the overhead increases as the amount of global state
information collected increases.
● This is because the usefulness of that information is decreased due to
both the aging of the information being gathered and the low
scheduling frequency as a result of the cost of gathering and
processing the extra information.

Stability:
● Fruitless migration of processes, known as processor thrashing, must
be prevented.
E.g. if nodes n1 and n2 observe that node n3 is idle and then offload a
portion of their work to n3 without being aware of the offloading
decision made by the other node.
● Now if n3 becomes overloaded due to this it may again start
transferring its processes to other nodes.
● This is caused by scheduling decisions being made at each node
independently of decisions made by other nodes.

79
Distributed System and 6.3.2 Local Scheduling
Cloud Computing
In a distributed system, local scheduling means how an individual
workstation should schedule those processes assigned to it in order to
maximize the overall performance. It seems that local scheduling is the
same as the scheduling approach on a stand-alone workstation. However,
they are different in many aspects. In a distributed system, the local
scheduler may need global information from other workstations to achieve
the optimal overall performance of the entire system. For example, in the
extended stride scheduling of clusters, the local schedulers need global
ticket information in order to achieve fairness across all the processes in
the system.
In recent years, there have been many scheduling techniques developed in
different models. Here, we introduce two of them: one is a proportional-
sharing scheduling approach, in which the resource consumption rights of
each active process are proportional to the relative shares that it is
allocated. The other is predictive scheduling, which is adaptive to the CPU
load and resource distribution of the distributed system.
The traditional priority-based schedulers are difficult to understand and
give more processing time to users with many jobs, which leads to
unfairness among users. Numerous researches have been trying to find a
scheduler that is easy to implement and can solve the problem of
allocating resources to users fairly over time. In this environment,
proportional-share scheduling was brought out to effectively solve this
problem. With proportional-share scheduling, the resource consumption
rights of each active process are proportional to the relative shares that it is
allocated.

6.3.3 Stride Scheduling:

As a kind of proportional-share scheduling strategies, stride scheduling
allocates resources to competing users in proportion to the number of
tickets they hold. Each user has a time interval, or stride, inversely
proportional to his/her ticket allocation, which determines how frequently
it is used. A pass is associated with each user. The user with a minimum
pass is scheduled at each interval; a pass is then incremented by the job's
stride.

6.3.3.1 Extension to Stride Scheduling:

The original stride scheduling only deals with CPU-bound jobs. If the
proportional-share schedulers are to handle the interactive and I/O
intensive job workloads, they must be extended to improve the responsive
time and I/O throughput, while not penalizing competing users. Here we
discuss two extensions to stride scheduling that give credits to jobs not
competing for resources. In this way, jobs are given incentive to relinquish
the processor when not in use and will receive their share of resources
over a longer time-interval. Thus, because interactive jobs are scheduled
more frequently when they awaken, they can receive better response time.
The first approach is loan & borrow, and the second approach is system
80
80
credit. Both approaches are built upon exhaustible tickets, which are Distributed System
simple tickets with expiration time. Management

Loan & Borrow: In this approach, exhausted tickets are traded among
competing clients. When a user temporarily exits the system, other users
can borrow these otherwise inactive tickets. The borrowed tickets expire
when the user rejoins the system. When the sleeping user wakes up, it
stops loaning tickets and is paid back in exhaustible tickets by the
borrowing users. In general, the lifetime of the exhaustible tickets is equal
to the length the original tickets were borrowed. This policy can keep the
total number of tickets in the system constant over time; thus, users can
accurately determine the amount of resources they receive. However, it
also introduces an excessive amount of computation into the scheduler on
every sleep and wake-up event, which we don’t expect.
System Credit: This second approach is an approximation of the first one.
With system credits, clients are given exhaustible tickets from the system
when they awaken. The idea behind this policy is that after a client sleeps
and awakens, the scheduler calculates the number of exhaustible tickets
for the clients to receive its proportional share over some longer interval.
The system credit policy is easy to implement and does not add significant
overhead to the scheduler on sleep and wakeup events.
Proportional-share of resources can be allocated to clients running
sequential jobs in a cluster. In the cluster, users are guaranteed a
proportional-share of resources if each local stride-scheduler is aware of
the number of tickets issued in its currency across the cluster and if the
total number of base tickets allocated on each workstation is balanced. The
solution for the first assumption is simple: each local scheduler is
informed of the number of tickets issued in each currency, and then
correctly calculates the base funding of each local job. The solution for
distributing tickets to the stride-schedulers is to run a user-level tickets-
sever on each of the nodes in the cluster. Each stride-scheduler
periodically contacts the local ticket server to update and determine the
value of currencies.
Further, for parallel jobs in a distributed cluster, proportional-share
resources can be provided through a combination of stride-scheduling and
implicit coscheduling. Preliminary simulations of implicit coscheduling
for a range of a communication patterns and computation granularity
indicate that the stride-scheduler with system credit performs similarly to
the Solaris time-sharing scheduler which is used in the Berkeley NOW
environment

6.3.4 Predictive Scheduling:

Predictive scheduling differs from other scheduling approaches in that it
provides intelligence, adaptivity and proactivity so that the system
implementing predictive scheduling can adapt to new architectures and/or
algorithms and/or environmental changes automatically. Predictive
scheduling can learn new architectures, algorithms and methods that are
embedded into the system. They provide some guarantees of service.
81
Distributed System and Furthermore, they are able to anticipate significant changes to its
Cloud Computing environment and avoid those changes to become the system performance
bottleneck.
Predictive scheduling can be roughly decomposed into three components:
H-cell, S-cell and allocator. The H-cell receives information of hardware
resource changes such as disk traffic, CPU usage, memory availability,
etc., and provides near-real-time control. Meanwhile, S-cell provides long-
term control of computational demands--such as what the deadline of a
task is and what its real-time requirement is--by interrogating the parallel
program code. H-cell and S-cell respectively collect information about
computational supply and computational demand, and provide to the
allocator the raw data or some intelligent recommendations. The allocator
reconciles the recommendations sent by the H-cells and S-cells and
schedules jobs according to their deadline, while guaranteeing constraints
and enforcing the deadline.
In the allocator, the previous inputs, in the form of a vector of
performance information (such as memory, CPU, disk usage etc.), are
aggregated into sets. Each set corresponds to a scheduling decision. The
allocator re-organizes the sets dynamically to keep a limited memory
demand by splitting or merging sets. If a new input matches one of the
pattern categories, a decision will be made due to the corresponding
decision of that pattern set, otherwise a new pattern category is built to
associate this new input pattern with corresponding scheduling decision.
Most of the scheduling policies are used either when a process blocks or at
the end of a time slice, which may reduce the performance because there
can be a considerable lapse of time before scheduling is done. Predictive
scheduling solves this problem by predicting when a scheduling decision
is necessary, or predicting the parameters needed by the scheduling
decision when not known in advance. Based on the collected static
information (machine type, CPU power, etc.) and dynamic information
(memory free space, CPU load, etc.), predictive scheduling tries to make
an educated guess about the future behavior, such as CPU idle time slot,
which can be used to make scheduling decisions in advance. Predicting the
future performance based on past information is a common strategy, and it
can achieve a satisfactory performance in practical work.
Predictive scheduling is very effective in performance and reliability
enhancement, even with the simplest methods, but at the cost of design
complexity and management overhead. Furthermore, it is observed that
the more complicated method is used, the more design complexity and
management overhead, and the less performance and reliability
enhancement.

6.3.5 Coscheduling:
In 1982, Outsterhout introduced the idea of coscheduling , which
schedules the interacting activities (i. e., processes) in a job so that all the
activities execute simultaneously on distinct workstations. It can produce
82
82 benefits in both system and individual job efficiency. Without coordinated
scheduling, the processor thrashing may lead to high communication Distributed System
latencies and consequently degraded overall performance. With systems Management
connected by highperformance networks that already achieve latencies
within tens microseconds, the success of coscheduling becomes a more
important factor in deciding the performance.

6.3.6 Gang Scheduling:

Gang scheduling is a typical coscheduling approach, which has already
been introduced for a long time but still plays a fundamental role.
Moreover, there are still many research projects in progress to improve
gang scheduling. The approach identifies a job as a gang and its
components as gang members. Further, each job is assigned to a class that
has the minimum number of workstations that meet the requirement of its
gang members based on a oneprocess- one-workstation policy. The class
has a local scheduler, which can have its own scheduling policy. When a
job is scheduled, each of its gang members is allocated to a distinct
workstation, and thus, the job executes in parallel. When a time-slice
finishes, all running gang members are preempted simultaneously, and all
processes from a second job are scheduled for the next time-slice. When a
job is rescheduled, effort is also made to run the same processes on the
same processors.
The strategy bypasses the busy-waiting problem by scheduling all
processes at the same time. According to the experience, it works well for
parallel jobs that have a lot of inter-process communications. However, it
also has several disadvantages. First, it is a centralized scheduling
strategy, with a single scheduler making decisions for all jobs and all
workstations. This centralized nature can easily become the bottleneck
when the load is heavy. Second, although this scheduler can achieve high
system efficiency on regular parallel applications, it has difficulty in
selecting alternate jobs run when processes block, requiring simultaneous
multi-context switches across the nodes. Third, to achieve good
performance requires long scheduling quanta, which can interfere with
interactive response, making them a less attractive choice for use in a
distributed system. These limitations motivate the integrated approaches.
The requirement of centralized control and the poor timesharing response
of previous scheduling approaches have motivated new, integrated
coscheduling approaches. Such approaches extend local timesharing
schedulers, preserving their interactive response and autonomy. Further,
such approaches do not need explicitly identified sets of processes to be
coscheduled, but rather integrate the detection of a coscheduling
requirement with actions to produce effective coscheduling.

6.3.7 Implicit Coscheduling:

Implicit coscheduling is a distributed algorithm for time-sharing
communicating processes in a cluster of workstations. By observing and
reacting to implicit information, local schedulers in the system make
independent decisions that dynamically coordinate the scheduling of
communicating processes. The principal mechanism involved is two-phase
83
Distributed System and spin-blocking: a process waiting for a message response spins for some
Cloud Computing amount of time, and then relinquishes the processor if the response does
not arrive.
The spin time before a process relinquishes the processor at each
communication event consists of three components. First, a process should
spin for the baseline time for the communication operation to complete;
this component keeps coordinated jobs in synchrony. Second, the process
should increase the spin time according to a local cost-benefit analysis of
spinning versus blocking. Third, the pairwise cost-benefit, i.e., the process,
should spin longer when receiving messages from other processes, thus
considering the impact of this process on others in the parallel job.
● The baseline time comprises the round-trip time of the network, the
overhead of sending and receiving messages, and the time to awake
the destination process when the request arrives.
● The local cost-benefit is the point at which the expected benefit of
relinquishing the processor exceeds the cost of being scheduled again.
For example, if the destination process will be scheduled later, it may
be beneficial to spin longer and avoid the cost of losing coordination
and being rescheduled later. On the other hand, when a large load
imbalance exists across processes in the parallel job, it may be
wasteful to spin for the entire load-imbalance even when all the
processes are coscheduled.
● The pairwise spin-time only occurs when other processes are sending
to the currently spinning process, and is therefore conditional.
Consider a pair of processes: the receiver who is performing a two-
phase spin-block while waiting for a communication operation to
complete, and a sender who is sending a request to the receiver. When
waiting for a remote operation, the process spins for the base and
local amount, while recording the number of incoming messages. If
the average interval between requests is sufficiently small, the process
assumes that it will remain beneficial in the future to be scheduled and
continues to spins for an additional spin time. The process continues
conditionally spinning for intervals of spin time until no messages are
received in an interval.

6.3.8 Dynamic Coscheduling:

Dynamic coscheduling makes scheduling decisions driven directly by the
message arrivals. When an arriving message is directed to a process that
isn’t running, a schedule decision is made. The idea derives from the
observation that only those communicating processes need to be
coscheduled. Therefore, it doesn’t require explicit identification to specify
the processes need coscheduling.

84
84
The implementation consists three parts: Distributed System
Management
Monitoring Communication/Thread Activity:
A firmware, which is on the network interface card, monitors the thread
activities by periodically reading the host's kernel memory. If the
incoming message is sent to the process currently running, the scheduler
should do nothing.

Causing Scheduling Decisions:

If a message received is not sent to the process currently running, an
interrupt will be produced and invoke the interrupt routine. When the
routine finds that it would be fair to preempt the process currently running,
the process receiving the message has its priority raised to the maximum
allowable priority for user mode timesharing processes, and is placed at
the front of the dispatcher queue. Flags are set to cause a scheduling
decision based on the new priorities. This will cause the process receiving
the message to be scheduled unless the process currently running has a
higher priority than the maximum allowable priority for user mode.

Making a Decision Whether to Preempt:

In dynamic coscheduling, the process receiving the message is scheduled
only if doing so would not cause unfair CPU allocation. The fairness is
implemented by limiting the frequency of priority boosts that therefore
limits the frequency of preemption. In jobs with fine-grained
communication, the sender and receiver are scheduled together and run
until one of them blocks or is preempted. Larger collections of
communicating processes are coscheduled by transitivity. The experiments
taken in HPVM project indicate that dynamic coscheduling can provide
good performance for a parallel process running on a cluster of
workstations in competition with serial processes. Performance was able
to close to ideal: CPU times were nearly the same as for batch processing,
and reduced job response times by up to 20% over implicit scheduling
while maintaining near-perfect fairness. Further, it claims that dynamic-
coscheduling-like approaches can be used to implement coordinated
resource management in a much broader range of cases, although most of
which are still to be explored.

6.4 TASK ASSIGNMENT APPROACH

Each process is viewed as a collection of tasks. These tasks are scheduled
to suitable processor to improve performance. This is not a widely used
approach because:
● It requires characteristics of all the processes to be known in advance.
● This approach does not take into consideration the dynamically
changing state of the system.
In this approach, a process is considered to be composed of multiple tasks
and the goal is to find an optimal assignment policy for the tasks of an
85
Distributed System and individual process. The following are typical assumptions for the task
Cloud Computing assignment approach:
● Minimize IPC cost (this problem can be modeled using network flow
model)
● Efficient resource utilization
● Quick turnaround time
● A high degree of parallelism
A Distributed System is a Network of Machines that can exchange
information with each other through Message-passing. It can be very
useful as it helps in resource sharing. In this article, we will see the
concept of the Task Assignment Approach in Distributed systems.

6.4.1 Resource Management:

One of the functions of system management in distributed systems is
Resource Management. When a user requests the execution of the process,
the resource manager performs the allocation of resources to the process
submitted by the user for execution. In addition, the resource manager
routes process to appropriate nodes (processors) based on assignments.
Multiple resources are available in the distributed system so there is a need
for system transparency for the user. There can be a logical or a physical
resource in the system. For example, data files in sharing mode, Central
Processing Unit (CPU), etc.
As the name implies, the task assignment approach is based on the
division of the process into multiple tasks. These tasks are assigned to
appropriate processors to improve performance and efficiency. This
approach has a major setback in that it needs prior knowledge about the
features of all the participating processes. Furthermore, it does not take
into account the dynamically changing state of the system. This
approach’s major objective is to allocate tasks of a single process in the
best possible manner as it is based on the division of tasks in a system. For
that, there is a need to identify the optimal policy for its implementation.

6.4.2 Working of Task Assignment Approach:

In the working of the Task Assignment Approach, the following are the
assumptions:
● The division of an individual process into tasks.
● Each task’s computing requirements and the performance in terms of
the speed of each processor are known.
● The cost incurred in the processing of each task performed on every
node of the system is known.

86
86
● The IPC (Inter-Process Communication) cost is known for every pair Distributed System
of tasks performed between nodes. Management

● Other limitations are also familiar, such as job resource requirements

and available resources at each node, task priority connections, and so
on.

6.4.3 Goals of Task Assignment Algorithms:

● Reducing Inter-Process Communication (IPC) Cost
● Quick Turnaround Time or Response Time for the whole process
● A high degree of Parallelism
● Utilization of System Resources in an effective manner
The above-mentioned goals time and again conflict. To exemplify, let us
consider the goal-1 using which all the tasks of a process need to be
allocated to a single node for reducing the Inter-Process Communication
(IPC) Cost. If we consider goal-4 which is based on the efficient
utilization of system resources that implies all the tasks of a process to be
divided and processed by appropriate nodes in a system.
Note: The possible number of assignments of tasks to nodes:
For m tasks and n nodes= m x n
But in practice, the possible number of assignments of tasks to nodes < m
x n because of the constraint for allocation of tasks to the appropriate
nodes in a system due to their particular requirements like memory space,
etc.

6.4.4 Need for Task Assignment in a Distributed System:

The need for task management in distributed systems was raised for
achieving the set performance goals. For that optimal assignments should
be carried out concerning cost and time functions such as task assignment
to minimize the total execution and communication costs, completion task
time, total cost of 3 (execution, communication, and interference), total
execution and communication costs with the limit imposed on the number
of tasks assigned to each processor, and a weighted product of cost
functions of total execution and communication costs and completion task
time. All these factors are countable in task allocation and turn, resulting
in the best outcome of the system.

6.4.5 Example of Task Assignment Approach:

Let us suppose, there are two nodes namely n1 and n2, and six tasks
namely t1, t2, t3, t4, t5, and t6. The two task assignment parameters are:
● execution cost: xab refers to the cost of executing a task an on node b.

87
Distributed System and ● inter-task communication cost: cij refers to inter-task
Cloud Computing communication cost between tasks i and j.

Task
s t1 t2 t3 t4 t5 t6

t1 0 6 4 0 0 12

t2 6 0 8 12 3 0

t3 4 8 0 0 11 0

t4 0 12 0 0 5 0

t5 0 3 11 5 0 0

t6 12 0 0 0 0 0

Execution Cost

Nodes
Task
s n1 n2

t1 5 10

t2 2 infinity

t3 4 4

t4 6 3

t5 5 2

t6 infinity 4

88
88
Note: The execution of the task (t2) on the node (n2) and the execution of Distributed System
the task (t6) on the node (n1) is not possible as it can be seen from the Management
above table of Execution costs that resources are not available.
Case1: Serial Assignment

Nod
Task e

t1 n1

t2 n1

t3 n1

t4 n2

t5 n2

t6 n2

Cost of Execution in Serial Assignment:

t11 + t21 + t31 + t42 + t52 + t62 = 5 + 2+ 4 + 3 + 2 + 4
= 20 (Refer Execution Cost table)

Cost of Communication in Serial Assignment:

= c14 + c15 + c16 + c24 + c25 + c26 + c34 + c35 + c36
= 0 + 0+ 12 + 12 + 3 + 0 + 0 + 11 + 0
= 38 (Refer Inter-task Communication Cost table)
Hence, Total Cost in Serial Assignment
= 20 + 38
= 58

89
Distributed System and Case2: Optimal Assignment
Cloud Computing
Task Node

t1 n1

t2 n1

t3 n1

t4 n1

t5 n1

t6 n2

Cost of Execution in Optimal Assignment:

= t11 + t21 + t31 + t41 + t51 + t62
= 5 + 2+ 4 + 6 + 5 + 4
= 26 (Refer Execution Cost table)

Cost of Communication in Optimal Assignment:

= c16 + c26 + c36 + c46 + c56
= 12 + 0+ 0 + 0 + 0
= 12 (Refer Inter-task Communication Cost table)
Hence, Total Cost in Optimal Assignment
= 26 + 12
= 38
6.5 Load balancing approach

90
90
6.5.1 Classification of Load Balancing Algorithms Distributed System
Management

Fig 5.11: Classification of Load Balancing Algorithms

Static versus Dynamic:

Static Algorithms Dynamic Algorithms

Static algorithms use only Dynamic algorithms collect state
information about the average information and react to system state
behavior of the system if it changed.
Static algorithms ignore the Dynamic algorithms are able to give
current state or load of the significantly better performance
nodes in the system.
Static algorithms are much They are complex
simpler

Deterministic versus Probabilistic:

Deterministic Algorithms Probabilistic Algorithms

Deterministic algorithms use Probabilistic algorithms use
the information about the information of static attributes of
properties of the nodes and the the system (e.g. number of nodes,
characteristic of processes to be processing capability, topology) to
scheduled. formulate simple process placement
rules
Deterministic approach is Probabilistic approach has poor
difficult to optimize. performance

Centralized versus Distributed:

Centralized Algorithms Distributed Algorithms

Centralized approach collects Distributed approach contains
information to server node and entities to make decisions on a
makes assignment decision predefined set of nodes
Centralized algorithms can make Distributed algorithms avoid the
efficient decisions, have lower bottleneck of collecting state
fault-tolerance information and react faster

91
Distributed System and Cooperative versus Non-cooperative:
Cloud Computing

Cooperative Algorithms Non-cooperative Algorithms

In Co-operative algorithms In Non-cooperative algorithms
distributed entities cooperate entities act as autonomous ones
with each other. and make scheduling decisions
independently from other entities.
Cooperative algorithms are They are simpler.
more
complex and involve larger
overhead
Stability of Cooperative Stability is comparatively poor.
algorithms are better.

6.5.2 Issues in Load Balancing Algorithms:

● Load estimation policy: determines how to estimate the workload of
a node.
● Process transfer policy: determines whether to execute a process
locally or remote.
● State information exchange policy: determines how to exchange
load information among nodes.
● Location policy: determines to which node the transferable process
should be sent.
● Priority assignment policy: determines the priority of execution of
local and remote processes.
● Migration limiting policy: determines the total number of times a
process can migrate.

6.6 LOAD-SHARING APPROACH

The following problems in load balancing approach led to load sharing
approach:
● Load balancing technique with attempting equalizing the workload on
all the nodes is not an appropriate object since big overhead is
generated by gathering exact state information.
● Load balancing is not achievable since number of processes in a node
is always fluctuating and temporal unbalance among the nodes exists
every moment

6.6.1 Basic idea:

It is necessary and sufficient to prevent nodes from being idle while some
92
92 other nodes have more than two processes.
● Load-sharing is much simpler than load-balancing since it only Distributed System
attempts to ensure that no node is idle when heavily node exists. Management

● Priority assignment policy and migration limiting policy are the same
as that for the load-balancing algorithms.

6.6.2 Load Estimation Policies:

Since load-sharing algorithms simply attempt to avoid idle nodes, it is
sufficient to know whether a node is busy or idle.
● Thus these algorithms normally employ the simplest load estimation
policy of counting the total number of processes.
● In modern systems where permanent existence of several processes on
an idle node is possible, algorithms measure CPU utilization to
estimate the load of a node

6.6.3 Process Transfer Policies:

● The load sharing algorithms normally use all-or-nothing strategy.
● This strategy uses the threshold value of all the nodes fixed to 1.
● Nodes become receiver node when it has no process, and become
sender node when it has more than 1 process.
To avoid processing power on nodes having zero process load-sharing
algorithms uses a threshold value of 2 instead of 1. When CPU utilization
is used as the load estimation policy, the double-threshold policy should
be used as the process transfer policy

Policies:
The location policy decides whether the sender node or the receiver node
of the process takes the initiative to search for suitable node in the system,
and this policy can one of the following:
Sender-initiated location policy: Sender node decides where to send the
process. Heavily loaded nodes search for lightly loaded nodes
Receiver-initiated location policy: Receiver node decides from where to
get the process. Lightly loaded nodes search for heavily loaded nodes
initiated location policy: Node becomes overloaded, it either broadcasts
or randomly probes the other nodes one by one to find a node that is able
to receive remote processes. When broadcasting, suitable node is known
as soon as reply arrives
initiated location policy: Nodes becomes underloaded, it either broadcast
or randomly probes the other nodes one by one to indicate its willingness
to receive remote processes. Receiver-initiated policy require preemptive
process migration facility since scheduling decisions are usually made at
process departure epochs

93
Distributed System and ● Both policies gives substantial performance advantages over the
Cloud Computing situation in which no load-sharing is attempted.
● Sender-initiated policy is preferable at light to moderate system loads.
● Receiver-initiated policy is preferable at high system loads.
● Sender-initiated policy provide better performance for the case when
process transfer cost significantly more at receiver-initiated than at
sender-initiated policy due to the pre-emptive transfer of processes.

information exchange policies:

In load-sharing algorithms it is not necessary for the nodes to periodically
exchange state information, but needs to know the state of other nodes
when it is either underloaded or overloaded. The following are the two
approaches followed when there is state change:

Broadcast when state changes:

● In sender-initiated/receiver-initiated location policy a node broadcasts
State Information Request when it becomes overloaded/ underloaded.
● It is called broadcast-when-idle policy when receiver-initiated policy
is used with fixed threshold value of 1

Poll when state changes:

● In large networks polling mechanism is used.
● Polling mechanism randomly asks different nodes for state
information until find an appropriate one or probe limit is reached.
● It is called poll-when-idle policy when receiver-initiated policy is
used with fixed threshold value value of 1.

6.6.4 Differences between Load Balancing and Load Sharing:

Load Balancing Load Sharing

1. Load balancing equally Load sharing delivers a portion of
distributes network traffic or the traffic or load to one
load across different channels connection in the network while
and can be achieved using both the remainder is routed through
static and dynamic load other channels.
balancing techniques.
2. Focuses on the notion of traffic Works with the notion of traffic
dispersion across connections. splitting across connections.
3. The creation of Ratios, Least Load Sharing is based on the
connections, Fastest, Round notion of sharing traffic or network
robin, and observed approaches load among connections based on
are used in load balancing. destination IP or MAC address
selections.
4. It is Uni-Directional. It is Uni-Directional
94
94
5. No instance is load sharing. All instances are load sharing. Distributed System
6. Accurate Load Balancing is not Load sharing is easy compared Management
an easy task. with load balancing.

6.7 SUMMARY
A distributed file system (DFS) is a network file system wherein the file
system is distributed across multiple servers. DFS enables location
transparency and file directory replication as well as tolerance to faults.
Some implementations may also cache recently accessed disk blocks for
improved performance. Though distribution of file content increases
performance considerably, efficient management of metadata is crucial for
overall file system performance. It has been shown that 75% of all file
system calls access file metadata [15] and distributing metadata load is
important for scalability. Scaling metadata performance is more complex
than scaling raw I/O performance since even a small inconsistency in
metadata can lead to data corruption.
Resource management is the process by which businesses manage their
various resources effectively. Those resources can be intangible – people
and time – and tangible – equipment, materials, and finances.
It involves planning so that the right resources are assigned to the right
tasks. Managing resources involves schedules and budgets for people,
projects, equipment, and supplies. While it is often used in reference to
project management, it applies to many other areas of business
management. A small business, in particular, will pay attention to resource
management in a number of areas

6.8 REFERENCE FOR FURTHER READING

1. https://www.shopify.in/encyclopedia/resource-management
2. https://www.wrike.com/blog/what-is-resource-
management/#:~:text=Resource%20management%20is%20the%20pr
ocess,or%20the%20adoption%20of%20software.
3. https://www.scality.com/topics/what-is-a-distributed-file-system
4. https://www.geeksforgeeks.org/what-is-dfsdistributed-file-system/
5. https://www.weka.io/learn/distributed-file-system/
6. https://www.techopedia.com/definition/1825/distributed-file-system-
dfs

6.9 MODEL QUESTIONS

1. Discuss on Types of Distributed Systems
2 Explain Advantages and disadvantages of Distributed Systems

95
Distributed System and 3. How does a distributed system work?
Cloud Computing
4. Explain the Architecture of resource Management in distributed
Environment
5. Write a short note on Features of scheduling algorithms
6. Explain any two scheduling algorithms in detail
7. Explain Resource Management
8. Discuss on Working of Task Assignment Approach
9. Explain the Goals of Task Assignment Algorithms
10. Discuss the Need for Task Assignment in a Distributed System
11. Explain with example the Task Assignment Approach
12. Explain the concept of Classification of Load Balancing Algorithms
13. What are the Issues in Load Balancing Algorithms
14. Explain the Basic idea of load sharing
15. Differentiate between Load Balancing and Load Sharing

*****

96
96
 7
DISTRIBUTED SYSTEM MANAGEMENT
Unit Structure
7.1 Process Management
7.1.1 Introduction
7.1.2 What is Process Management?
7.1.3 The Importance of Process Management
7.1.4 Realtime Process Management Examples
7.1.5 How does BPM differ from workflow management?
7.1.6 Distinction between Digital Process Automation (DPA) and
Process Management
7.1.7 Examples of Business Process Management (BPM) phases
7.1.8 Digital process management: Application examples
7.1.9 How can process management be implemented?
7.1.10 Selection criteria for good BPM software
7.1.11 Benefits of Process Management
7.2 Process Migration in Distributed System
7.2.1 Why use Process Migration?
7.2.2 What are the steps involved in Process Migration?
7.2.3 Methods of Process Migration
7.3 Threads
7.4 Summary
7.5 Reference for further reading
7.6 Model questions

7.1 PROCESS MANAGEMENT

7.1.1 Introduction:
All business organizations involve processes. It can be as simple as buying
ingredients, baking bread, selling bread, and receiving payment for a
bakery. It can also be more complex, like a multistep purchasing process
for vendor management. In either case, without an efficient system,
unorganized processes can lead to problems that can adversely affect a
business. Thus, it is important to implement process management
regardless of the size of the company. To understand its importance,
here’s a quick run-through of how process management
benefits business organizations.

7.1.2 What is Process Management?:

Process management is a systematic approach to ensure that effective and
efficient business processes are in place. It is a methodology used to align
97
Distributed System and business processes with strategic goals. In contrast to project management,
Cloud Computing which is focused on a single project, process management addresses
repetitive processes carried out on a regular basis. It looks at
every business process, individually and as a whole, to create a more
efficient organization. It analyzes current systems, spots bottlenecks, and
identifies areas of improvement. Process management is a long-term
strategy that constantly monitors business processes so they maintain
optimal efficiency. Implemented properly, it significantly helps boost
business growth.

7.1.3 The Importance of Process Management:

When managing any organization, it is imperative to understand why
process management is important. More than creating seamless
workflows, it enables all aspects of business operations to run at an
optimal pace. With business processes systematically implemented, you
reduce time wasted on repetitive tasks and minimize errors due to human
inefficiency. It also prevents the loss of data and missed steps within a
process. Moreover, it ensures that resources are used properly so your
business becomes more cost-efficient. Aside from improving business
operations, process management also aligns your processes with the needs
of your customers. This increases customer satisfaction and leads to higher
revenues.

7.1.4 Realtime Process Management Examples:

Process management encompasses all aspects of business. Some business
organizations use process management software to automate their systems,
while others still use traditional methods of flowcharts and manuals. If you
have been running your own company, you are most likely using some
form of business process management (BPM).
Here are some real-world examples of how BPM is used in various
industries.
Onboarding New Employees | Human Resource Department:
Without a proper system, the onboarding process can be chaotic and time-
consuming. With BPM, forms and documents can be filled up and
submitted electronically. Software is used to automatically filter data, find
the best matches for a position, send messages, schedule interviews, and
facilitate employee onboarding.

Managing Logistics | Shipping Company:

Logistics for a shipping company entails a long chain of complex
processes while dealing with potentially thousands of people in various
locations. BPM standardizes and optimizes routines involved to streamline
the entire process and deliver quality service. It integrates production,
finance, quality control, HR, customer service, and other departments. It
centralizes data to facilitate easy retrieval of information in every phase of
the business operations.
98
98
Loan Processing | Banking Firms: Distributed System
Management
With BPM, processing loans can be done in a much shorter time. It creates
an efficient flow from document submission to credit and risk checks to
loan approval. It also enables tracking of applications through the entire
loan processing system.

Compliance Management| Insurance Companies:

BPM improves the overall regulatory compliance of insurance companies.

It reduces human error and prevents data loss through proper
documentation management. It also ensures that the company is able to
comply with the latest state and federal regulations.

Customer Service | Retail Business:

BPM enables customer-centric operations. It unifies all systems and

departments for a smooth workflow that ensures all customer needs are
met. It also identifies bottlenecks in the buyer’s journey so the entire
purchasing process can be improved.

7.1.5 How does BPM differ from workflow management?:

Workflow management is another term that is often mentioned in

connection with process management. While workflow management is
about coordinating and organizing processes, process management is
about integrating these individual business processes into a larger whole.
Workflow management is therefore a subfield, but by no means a
synonym of process management. However, the aim of efficient processes
is the same in both approaches.

7.1.6 Distinction between Digital Process Automation (DPA) and

Process Management:

Digital Process Automation (DPA) is the next evolution in process

management and supports companies in their efforts towards end-to-end
digital transformation. In contrast to DPA, process management initially
has nothing to do with digital transformation. However, in order to
achieve the desired improvements, digital transformation usually involves
companies digitizing and ultimately automating processes. In the case of
DPA, company-wide business processes with external participants come
more into focus. DPA is a technology that ensures that processes and
systems are not only digitized, but that tasks and sequences that would
normally require manual intervention are automated.

99
Distributed System and
Cloud Computing

Examples of Business Process Management (BPM) phases

Analysis:
In the analysis phase, processes are first identified and then analyzed
(people, activities, individual steps, points in time, etc.).

Modeling:
The modeling phase essentially involves selecting and adapting processes
that are to be implemented.

Execution:
Once processes are defined, the execution phase begins, including efforts
to automate business processes.

Monitoring:
The monitoring phase is the prerequisite for subsequent optimization and
is used for targeted process control.

Optimization:
The optimization phase begins following monitoring. The knowledge
gained during monitoring makes it possible to improve the processes. It is
possible, for example, that there are subtasks that have not been
automated, that unnecessary steps are still being carried out, or that the
data structure in general may require readjustment.

7.1.8 Digital process management: Application examples:

Throughout a company or in individual departments, there are various
workflows that contain individual steps, as well as instructions for actions
and responsibilities. These processes can be both internal and external.
100
100 Examples of internal processes are the hiring of new employees or the
ordering of office supplies. Invoicing processes (processing accounts Distributed System
payable invoices or creating invoices), on the other hand, are external Management
processes because outsiders (customers, partners or suppliers) are
involved.

Use cases that affect the entire company are, for example:
● Accounting and finance
● Purchasing decisions
● Administrative activities
● Customer services
● Facility management
● Personnel management
● Order processing
● Performance measurement
● Warehousing, logistics
● Standard operating procedures
● Employee performance and training
● Supplier and customer portal
In addition, there are processes whose origin can be assigned to a specific
department. All these workflows can be mapped digitally. Even if the
actual process management is detached from digital solutions, it can be
implemented much more easily. Digital and automated business processes
improve performance in all departments - this reduces overhead and
enables flexibility in the company.

Human Resources:
With the advancement of organization-wide digital strategies, the HR
department is also changing - and gaining in importance. In addition to
administrative tasks such as approving vacation requests or checking
applications, the HR department also has a strategic role to play in the
"war for talent". In the future, fewer administrative tasks will be on the
agenda than creative, intelligent activities in the corporate environment.
This, in turn, requires companies to digitize and automate HR processes
that used to be time-consuming and to connect them with the relevant data
and departments. More flexible work structures, further developing
corporate culture and supporting employees in digital transformation and
change management are today's new priorities. To free up time for
strategic and value-creating activities, HR employees must be relieved of
their administrative tasks. This is done through digital business processes.
Here are some workflows that can be digitized and automated in HR:

101
Distributed System and ● Vacation request
Cloud Computing
● Digital personnel file
● Travel expense report
● Application management
● Employee onboarding
● Expense report
● Business trip application
And here are the reasons why digital automaton is particularly worthwhile
in the HR sector:
● Promotion of new and agile ways of working and mobile working
● Increased employee satisfaction through fast processing
● Transparency in employee responsibilities
● Simplified change management
● Information exchange without interruption or integration problems

Administration:
Who hasn't had to handle a contract or obtain a permit? These tasks are
often time-consuming when performed in analog (manual) form.
Digitizing administrative processes is therefore an essential part of any
digitization strategy. In addition, workflows can only be accelerated
effectively when all systems, applications and company departments are
connected locally and across decentralized locations. To achieve this,
companies have to rethink their analog processes, sometimes redesigning
and streamlining them. As a result, the heterogeneous IT landscape
becomes interconnected, the company remains agile, and decision-makers
can focus on management tasks. This frees up employees to use their time
more effectively - and allows the company to maintain control over all
physical and electronic records at all times. Additionally, the newly
created visibility helps identify operational bottlenecks and continuously
improve processes. Here are a few workflows that can be digitized and
automated in administration:
● Contract management
● Training management
● Approval processes
● Digital construction file / digital project file
● Maintenance order / Production order

102
102 ● Fleet management
And these are the reasons why digital automation in administration is Distributed System
worthwhile: Management

● Enormous time savings across all departments

● Faster and more up-to-date accounting
● Faster, more secure access to all documents and data
● Efficient workflows thanks to end-to-end processes
● Optimized service thanks to prompt, proactive contact
● Real-time daily accounting and payroll reports
● Reduced overall process costs
● Well-founded decisions thanks to increased transparency
● Strict security standards for documents and data exchange

Finance:
The finance department is undergoing a transformation. Administrative
tasks still fill the business day, but according to McKinsey Research, as
much as 75-79 percent of general accounting operations, cash
dispurement, and revenue management tasks in finance can be fully
automated in the future. Experts are convinced that with digitization,
employees in finance departments are developing into business partners,
answering trend-setting questions, interpreting data and contributing
increasing value to their organization. The problem arises when the
number of cross-departmental, control-relevant data continues to increase
- but beneficial processes for management are not implemented. As real-
time and ad-hoc analysis grows in importance along the entire value chain,
it will become a priority to set up an interconnected value-creation system.
Here are some workflows that can be digitized and automated in finance:
● Treasury Management
● Risk Assessment
● Purchase-to-Pay
o Invoice Receipt Verification
o Outgoing invoice processing
o Payment processing
● Expenditure planning
● bank report
● data controlling
● Accounts Payable / Accounts Receivable
103
Distributed System and And these are the reasons why digital automation in finance is worthwhile:
Cloud Computing
● More performance through integrated data exchange
● Creation of electronic invoices with information linking
● Establishment of a paperless filing system as a central reference for
all documents
● Ability to exchange, match and archive documents without material
costs
● Optimized cost control

Purchasing:
It is no secret that digital procurement processes can now be fully
automated. Companies also rely on operational and in-house digital
processes such as requirements gathering and pricing, with data from
various sources being integrated. However, it is the comprehensive
exchange of information that brings the full advantage of digital
procurement to light. Sustainable processes require more than strategic
data management. The degree of interconnection between employees,
departments and systems determines how digital and efficient purchasing
can be. In day-to-day business, this can be seen by optimizing the supply
chain and maximizing response time. With digital processes, the modern
buyer maintains full control and transparency over processes, tasks and
figures at all times, and can make decisions in real time, despite the large
number of purchasing processes involved.
Here are some workflows that can be digitized and automated in
purchasing:
● Investment request
● Goods receipt process
● Order processing
● Inventory process
● Delivery release
And these are the reasons why digital automation in purchasing is
worthwhile:
● Contract, supplier and procurement management with seamless
system and data integration
● Optimized supply chains
● Maximized reaction speed
● Automated routine processes (article dispositions, creation of order
proposals or price inquiries)
104
104
● Transparent bookings and stock levels Distributed System
Management
● New savings potential
● Reduced processing time

Sales:
Today, more than 50 percent of new employees already belong to the
"digital native" generation. They have has grown up with digital tools and
different ways of working and are transferring this experience to their
everyday working lives. They don't think much of mountains of
documents and Excel lists with manually prepared data or paperwork.
Customers have also opened up new information channels. By the time the
first contact with sales is made, the decision has often already been made,
so advance work must be done - on all channels. After all, the customer
should have the choice of how to get in touch and the employee should be
able to switch seamlessly between channels to qualify a lead efficiently.
Information overload and attention deficits among prospects demand that
companies streamline all sales processes and optimize them digitally.
Information must be quick and easy to find, and data must be efficient to
use. Here are some workflows that can be digitized and automated
in sales:
● Order processing
● Information download
● Quotation approval
● Compilation of product sheets
● B2B sales process
And these are the reasons why digital automation in sales is worthwhile:
● Simplification of processes, communication and advice
● Increased reach and sales
● Optimized sales productivity
● Build trust and prevent mistrust
● Reduced costs for administration and organization
● Increased effectiveness and reduced susceptibility to errors
● Simplified contact, data maintenance & collaboration
● Sustainable competitive advantage

HR, finance and accounting are high on the digital agenda:

Recent surveys show that companies reported they are gaining momentum
digitally, especially in HR, finance and accounting, but are still far from
105
Distributed System and exploiting all the possibilities. Companies that want to introduce process
Cloud Computing management should pursue a holistic strategy while taking small steps
forward in practice. Time will tell if other areas should wait until later in
the journey and therefore lag behind digitally.
In particular, HR, finance, and accounts payable topics are digitized:
In which of the following areas does your company already use or plan to
use digital solutions for planning and controlling company resources?

7.1.9 How can process management be implemented?:

How process management works can be explained in theory with the BPM
lifecycle. But theory and practice are often only comparable to a limited
extent. In practice, you should be aware that process management is not a
rigid process of working through to-do lists, but has a lot to do with
communication and teamwork. Pilot projects for digital automation in
companies often start in isolation from one another. This silo approach can
result in projects that become stalled, terminate unsuccessfully, or create
islands of information that impede information flow to other teams across
the organization. Surveys by Accenture and McKinsey indicate that when
companies approach digital process management holistically from the
outset with the necessary communication, failure can be prevented. It is
also important to have the backing of all stakeholders, including
management, in order to introduce and successfully drive digital process
management forward to its full potential.
Learn about three essential steps to implement process management here.

1. Organization-wide business process structure:

Once you have gained support in the company for this topic, the first step
is to get an overview of the external and internal processes in your
company. Talk to colleagues from other departments and outline all
workflows. Ideally, you should already record all the processes that take
place in the company. Alternatively, you can take a department-specific
approach and, for example, first analyze business processes in human
resources or administration. It is useful to know that there are different
types of processes that fulfill different tasks and purposes and ultimately
help you to structure business processes. As a rule, value creation, support
and management processes are distinguished from one another.

106
106
Value creation processes: Distributed System
Management
Value-adding processes are essential for the creation of a product or the
provision of a service. They describe all corporate activities that are
geared to customer needs. The value-adding processes a company has
depend to a large extent on its industry focus or its core competencies.
Typically, sales and marketing processes are almost always part of the
value creation processes. It is also characteristic that different departments
of a company are integrated into the value creation processes.

Support processes:
Support processes, also known as supporting processes, are not customer-
oriented at first, but are necessary in order to carry out, control and
optimize value creation and management processes. These include, for
example, personnel selection and qualification as well as purchasing or the
payment of invoices. In contrast to value-added processes, support
processes can often be assigned to a single department.

Management processes:
Management processes relate to the company as a whole, contribute to the
planning and control of core and support processes, and serve to
strategically manage a company. Similar to support processes, this type of
process is not directly related to the value creation of a company.
Examples of management processes would include, but are not limited to:
Aligning the company strategically, defining the corporate mission
statement or formulating corporate goals.

2. Create process map:

The second step emerges from the discussions with internal and external
stakeholders. The process map graphically depicts all processes in the
company. Division into core, support and management processes is
supplemented by the interactions/dependencies of business processes with
each other. Figuratively speaking, the process map is your compass to
keep focus on the path to process management. In addition, the process
map can be consulted to explain the process management projects to
employees. This clear representation helps to ensure understanding of the
(change) project. In addition, a graphical representation makes it
particularly easy to identify optimization potential. You can create a
process map by enriching your core, support and management processes
with additional information (e.g., responsible department/employee) and
further subdividing them into sub-processes (if appropriate) and adding
required third-party systems, documents and key figures.

3. Implement and optimize processes:

Once an overview has been created, you can work through the business
processes from your map step-by-step, digitizing and even automating
them. With subsequent process optimization, the BPM lifecycle then
begins again.

107
Distributed System and Before you start with the implementation, however, you should have
Cloud Computing completed important tasks:
● Involve employees early on and offer training.
● Record process steps, responsibilities and other important information
in documentation.
● Identify and contact all stakeholders in a timely manner. Also
consider those who have no direct contact with the process but must
approve it - for example, the works council or data protection officers.
It is also advisable to set regular coordination and review dates even
before the introduction. In this way, you can ensure from the outset that
processes are constantly scrutinized and, if necessary, adapted or even
eliminated. We have also compiled five additional tips to help you
successfully implement your project management.

Five tips for successful process management

1. Understand the current process:
Let's take an ordering process as an example. Do you know who orders
what in your company and what approvals need to be obtained? No? Then
you are in the same situation as many others. That is why you should get
rid of the idea that you, or even company management, can have detailed
knowledge of all processes. From a practical point of view, the best people
to talk to are still those who work with the process on a regular basis.
Don't get too hung up on theory. Talking to colleagues will tell you about
individual process steps and everything else you need to know to get
started.

2. Start small:
"It's hard to get started" - a saying that couldn't be more apt for process
management. So don't make things unnecessarily difficult for yourself and
start small. Choose simple processes that address the current situation and
avoid being overwhelmed. Especially if process management is new. You
should aim for small stages that can be implemented quickly and easily.
Success will inspire you to do more - so you can gradually tackle other, as
yet untried processes in your company.

3. Create a schedule:
You will probably realize pretty quickly that many things will take much
longer than initially anticipated. Nevertheless, it makes sense to create a
schedule that includes generous milestones and that you can regularly
review and adjust. Process management is not something you can just do
on the side. It is important to give all involved employees sufficient time
to deal with the topic in detail and successfully implement their tasks.

108
108
4. Enable the exchange of information, ideas: Distributed System
Management
As with any new big project, sooner or later challenges crop up.
Restructuring, new service providers and much more can mean that you
have to throw your well-thought-out plan out the window and start over
again. Regular meetings are a good way to learn about changes early on
and to support each other. Bring yourself up to date, discuss problems and
inspire each other. The intervals at which you schedule meetings depend
on how much you need to discuss. It is advisable to choose shorter
intervals, especially during critical phases, such as at the beginning. Once
everything has settled in, weekly or monthly meetings are also advisable.

5. Use tools:
With your process map, you already have a powerful tool at hand. It is the
basis for analysis, meetings and further developments and should always
be included in the regular coordination meetings. But the choice of a
suitable means of communication is also crucial. E-mails often lead to
misunderstandings and ultimately cause more confusion than they help the
process. Therefore, only write e-mails if your request can be explained
succinctly and be sure to file any important documents in an agreed-upon
location. This is where project management tools have proven to be the
tool of choice, as they bundle all communication including documents,
schedules, and to-do lists, etc. There are a number of software products
available, some of which can even be used free of charge - for example,
Asana or Trello.

7.1.10 Selection criteria for good BPM software:

With the help of BPM software, companies can design, implement,
optimize and, above all, automate business processes. All tasks are
performed in one platform. Accordingly, a BPM tool helps to approach
process management holistically and to view and improve processes from
start to finish. As a result, processes can not only be made more efficient,
but resources can be adjusted according to need, errors can be reduced,
time can be saved, and ultimately the entire value creation of a company
can be increased.
In order to take full advantage of this, it is important to choose a suitable
software. We have collected some criteria that experience has shown to
play an important role in the search for and selection of good BPM
software.

Monitoring:
Make sure that you can use the BPM software to monitor key business
indicators in real time, if possible. Ideally, you should be able to visualize
the data in a dashboard.

Scalability:
Every company has its own special features and unique requirements.
Therefore, make sure that the selected process management software can
109
Distributed System and handle them. This includes integration to third-party systems as well as
Cloud Computing preferred data types or archiving and search functions for documents. A
scalable solution is also necessary for companies with growth and
expansion plan. This enables them to handle not just current requirements,
but also future changes as well.

Security:
Recent surveys across many industries shows that data privacy and
security have become top priorities driving digital transformation
initiatives. Security is therefore one of the most important criteria when it
comes to selecting a suitable BPM tool. Particularly in the case of cloud
solutions where the question of data residency, where data is stored, is at
the forefront. In many countries and states, government regulations can be
particularly strict. It is therefore advisable to choose a software provider
that offers cloud options that meet data residency requirements. In
addition, process management software should meet the requirements of
data protection regulations such as GDPR.

Analysis and simulation:

You should make sure that the software provides an analysis and
simulation function. With a heat map, for example, it is possible to
analyze weak points within a process. The simulation function allows you
to test the process from the end user's point of view. This allows you to
check the functionality of subsequent steps, system activities, control
conditions or scripts and to detect errors in advance.

Usability:
For process designers, a flexible and user-friendly interface is essential.
This enables them to model and adapt even complicated process forms
quickly and easily. Of course, end users also benefit from user-friendly
BPM software that makes it particularly easy to start and control
processes.

7.1.11 Benefits of Process Management:

Streamlined Processes: BPM restructures tangled operations into smooth
workflows, simplifying operations and improving business agility.
Increased Productivity: BPM makes sure that resources and capital are
utilized properly. It also improves business processes and working
conditions to increase overall productivity.
Minimized Risks: By clearly defining responsibilities, BPM demands
higher accountability. This minimizes risks due to human error and
reduces inefficiencies.
Reduced Costs: BPM helps spot inefficiencies so they can be corrected. It
also tracks the usage of resources. With fewer inefficiencies and proper
utilization of resources, BPM can reduce costs and expenditures.
110
110
Business Process Management (BPM) is a method for analyzing, Distributed System
designing, controlling and ultimately improving business processes. Management
Ideally, all of a company's business processes are included in the analysis.
These can include, for example, processes with other companies, systems,
customers, suppliers or partners. The goal is to improve business processes
in such a way that they contribute optimally to achieving the company's
goals. Process management or business process management are the terms
for BPM and are used synonymously. Generally speaking, process
management includes both analog and digital processes.

In summary, process management ensures

● transparent costs and responsibilities,
● efficient cross-departmental processes,
Information and knowledge exchange between different departments.
Depending on the circumstances, the individual phases vary, but they
usually include a modeling, execution and monitoring phase

7.2 PROCESS MIGRATION IN DISTRIBUTED SYSTEM

A process is essentially a program in execution. The execution of a
process should advance in a sequential design. A process is characterized
as an entity that addresses the essential unit of work to be executed in the
system. A process is characterized as an entity that addresses the essential
unit of work to be executed in the system.
Process migration is a particular type of process management by which
processes are moved starting with one computing environment then onto
the next.

There are two types of Process Migration:

● Non-preemptive process: If a process is moved before it begins
execution on its source node which is known as a non-preemptive
process.
● Preemptive process: If a process is moved at the time of its
execution that is known as preemptive process migration. Preemptive
111
Distributed System and process migration is all the more expensive in comparison to the non-
Cloud Computing preemptive on the grounds that the process environment should go
with the process to its new node.

7.2.1 Why use Process Migration?:

The reason to use process migration are:
● Dynamic Load Balancing: It permits processes to exploit less
stacked nodes by relocating from overloaded ones.
● Accessibility: Processes that inhibit defective nodes can be moved to
other perfect nodes.
● System Administration: Processes that inhabit a node if it is going
through system maintenance can be moved to different nodes.
● The locality of data: Processes can exploit the region of information
or other extraordinary abilities of a specific node.
● Mobility: Processes can be relocated from a hand-operated device or
computer to an automatic server-based computer before the device
gets detached from the network.
● Recovery of faults: The component to stop, transport and resume a
process is actually valuable to support in recovering the fault in
applications that are based on transactions.

7.2.2 What are the steps involved in Process Migration?

The steps which are involved in migrating the process are:
● The process is chosen for migration.
● Choose the destination node for the process to be relocated.
● Move the chosen process to the destination node.

The subcategories to migrate a process are:

● The process is halted on its source node and is restarted on its
destination node.
● The address space of the process is transferred from its source node to
its destination node.
● Message forwarding is implied for the transferred process.
● Managing the communication between collaborating processes that
have been isolated because of process migration.

112
112
7.2.3 Methods of Process Migration: Distributed System
Management
The methods of Process Migration are:
1. Homogeneous Process Migration: Homogeneous process migration
implies relocating a process in a homogeneous environment where all
systems have a similar operating system as well as architecture. There are
two unique strategies for performing process migration. These are i) User-
level process migration ii) Kernel level process migration.
● User-level process migration: In this procedure, process migration is
managed without converting the operating system kernel. User-level
migration executions are more simple to create and handle but have
usually two issues: i) Kernel state is not accessible by them. ii) They
should cross the kernel limit utilizing kernel demands which are slow
and expensive.
● Kernel level process migration: In this procedure, process migration
is finished by adjusting the operating system kernel. Accordingly,
process migration will become more simple and more proficient. This
facility permits the migration process to be done faster and relocate
more types of processes.
Homogeneous Process Migration Algorithms:
There are five fundamental calculations for homogeneous process
migration:
● Total Copy Algorithm
● Pre-Copy Algorithm
● Demand Page Algorithm
● File Server Algorithm
● Freeze Free Algorithm

2. Heterogeneous Process Migration:

Heterogeneous process migration is the relocation of the process across
machine architectures and operating systems. Clearly, it is more complex
than the homogeneous case since it should review the machine and
operating designs and attributes, as well as send similar data as
homogeneous process migration including process state, address space,
file, and correspondence data. Heterogeneous process migration is
particularly appropriate in the portable environment where is almost
certain that the portable unit and the base help station will be different
machine types. It would be attractive to relocate a process from the
versatile unit to the base station as well as the other way around during
calculation. In most cases, this couldn’t be accomplished by homogeneous
migration. There are four essential types of heterogeneous migration:
● Passive object: The information is moved and should be translated
113
Distributed System and ● Active object, move when inactive: The process is relocated at the
Cloud Computing point when it isn’t executing. The code exists in the two areas, and
just the information is moved and translated.
● Active object, interpreted code: The process is executing through an
interpreter so just information and interpreter state need be moved.
● Active object, native code: Both code and information should be
translated as they are accumulated for a particular architecture.

7.3 THREADS
A thread is a light weight process which is similar to a process where
every process can have one or more threads. Each thread contains a Stack
and a Thread Control Block. There are four basic thread models :
1. User Level Single Thread Model:
● Each process contains a single thread.
● Single process is itself a single thread.
● process table contains an entry for every process by maintaining its
PCB.

2. User Level Multi Thread Model:

● Each process contains multiple threads.
● All threads of the process are scheduled by a thread library at user
level.
● Thread switching can be done faster than process switching.
● Thread switching is independent of operating system which can be
done within a process.
● Blocking one thread makes blocking of entire process.
● Thread table maintains Thread Control Block of each thread of a
process.
● Thread scheduling happens within a process and not known to Kernel.
114
114
 Distributed System
Management

3. Kernel Level Single Thread Model:

● Each process contains a single thread.
● Thread used here is kernel level thread.
● Process table works as thread table.

4. Kernel Level Multi Thread Model:

● Thread scheduling is done at kernel level.
● Fine grain scheduling is done on a thread basis.
● If a thread blocks, another thread can be scheduled without blocking
the whole process.
● Thread scheduling at Kernel process is slower compared to user level
thread scheduling.
● Thread switching involves switch.

115
Distributed System and
Cloud Computing

7.4 SUMMARY
A thread is also known as lightweight process. The idea is to achieve
parallelism by dividing a process into multiple threads. For example, in a
browser, multiple tabs can be different threads. MS Word uses multiple
threads: one thread to format the text, another thread to process inputs, etc.
The primary difference is that threads within the same process run in a
shared memory space, while processes run in separate memory spaces.
Threads are not independent of one another like processes are, and as a
result threads share with other threads their code section, data section, and
OS resources (like open files and signals). But, like process, a thread has
its own program counter (PC), register set, and stack space

7.5 REFERENCE FOR FURTHER READING

1. https://www.geeksforgeeks.org/thread-in-operating-system/
2. https://www.geeksforgeeks.org/threads-and-its-types-in-operating-
system/
3. https://www.javatpoint.com/threads-in-operating-system
4.
https://www.cs.uic.edu/~jbell/CourseNotes/OperatingSystems/4_Thre
ads.html
5. https://en.wikipedia.org/wiki/Thread_(computing)
6. https://padakuu.com/threads-summary-184-article
7.
https://www.tutorialspoint.com/operating_system/os_multi_threading.
htm
8. https://www.studytonight.com/operating-system/multithreading
116
116
7.6 MODEL QUESTIONS Distributed System
Management
1. What is Process Management?
2. Explain the Importance of Process Management?
3. Explain Realtime Process Management with Examples
4. How does BPM differ from workflow management?
5. Give the distinction between Digital Process Automation (DPA) and
Process Management ?
7. Explain Business Process Management (BPM) phases with
examples?
8. Explain application of Digital process management with examples?
9. How can process management be implemented?
10. What are Selection criteria for good BPM software?
11. Discuss the Benefits of Process Management ?
12. Why use Process Migration?
13. What are the steps involved in Process Migration?
14. Discuss the methods of Process Migration?
15. Explain Threads?

*****

117
 8
DISTRIBUTED SYSTEM MANAGEMENT
Unit Structure
8.1 Data file system
8.1 .1 What is DFS (Distributed File System)?
8.1.2 Features of DFS
8.1.3 History
8.1.4 Applications
8.1.5 Working of DFS
8.1.6 Advantages
8.1.7 Disadvantages
8.1.8 Benefits of DFS Models
8.1.9 DFS and Backup
8.1.10 The challenges associated with DFS
8.1.11 Components
8.2 File Models in Distributed System
8.3 Summary
8.4 Reference for further reading
8.5 Model questions

8.1 DATA FILE SYSTEM

A Distributed File System (DFS) is a file system that is distributed on
multiple file servers or multiple locations. It makes the programs to access
or to store isolated files with the local ones, allowing programmers to
access files from any network or computer. It manages files and folders on
different computers. It is mainly designed to provide file storage and
access controlled to files over LAN and WAN.
A DFS is also called a client-server architecture based application, which
allows the user or clients to access the data from the server as it is stored
in their own computer. It provides location transparency and redundancy
help to improve the data availability. And also use data replication strategy
on multiple servers to prevent data access failure.

8.1.1 What is DFS (Distributed File System)?:

A Distributed File System (DFS) as the name suggests, is a file system
that is distributed on multiple file servers or multiple locations. It allows
programs to access or store isolated files as they do with the local ones,
allowing programmers to access files from any network or computer.
The main purpose of the Distributed File System (DFS) is to allows users
of physically distributed systems to share their data and resources by using

118
a Common File System. A collection of workstations and mainframes Distributed System
connected by a Local Area Network (LAN) is a configuration on Management
Distributed File System. A DFS is executed as a part of the operating
system. In DFS, a namespace is created and this process is transparent for
the clients.

DFS has two components:

Location Transparency:
Location Transparency achieves through the namespace component.

Redundancy:
Redundancy is done through a file replication component.
In the case of failure and heavy load, these components together improve
data availability by allowing the sharing of data in different locations to be
logically grouped under one folder, which is known as the ―DFS root‖.
It is not necessary to use both the two components of DFS together, it is
possible to use the namespace component without using the file replication
component and it is perfectly possible to use the file replication
component without using the namespace component between servers.

8.1.2 Features of DFS:

Transparency:

Structure transparency:
There is no need for the client to know about the number or locations of
file servers and the storage devices. Multiple file servers should be
provided for performance, adaptability, and dependability.

Access transparency:
Both local and remote files should be accessible in the same manner. The
file system should be automatically located on the accessed file and send it
to the client’s side.

Naming transparency:
There should not be any hint in the name of the file to the location of the
file. Once a name is given to the file, it should not be changed during
transferring from one node to another.

Replication transparency:
If a file is copied on multiple nodes, both the copies of the file and their
locations should be hidden from one node to another.

119
Distributed System and User mobility:
Cloud Computing
It will automatically bring the user’s home directory to the node where the
user logs in.
Performance:
Performance is based on the average amount of time needed to convince
the client requests. This time covers the CPU time + time taken to access
secondary storage + network access time. It is advisable that the
performance of the Distributed File System be similar to that of a
centralized file system.
Simplicity and ease of use:
The user interface of a file system should be simple and the number of
commands in the file should be small.
High availability:
A Distributed File System should be able to continue in case of any partial
failures like a link failure, a node failure, or a storage drive crash.
A high authentic and adaptable distributed file system should have
different and independent file servers for controlling different and
independent storage devices.

Scalability:
Since growing the network by adding new machines or joining two
networks together is routine, the distributed system will inevitably grow
over time. As a result, a good distributed file system should be built to
scale quickly as the number of nodes and users in the system grows.
Service should not be substantially disrupted as the number of nodes and
users grows.

High reliability:
The likelihood of data loss should be minimized as much as feasible in a
suitable distributed file system. That is, because of the system’s
unreliability, users should not feel forced to make backup copies of their
files. Rather, a file system should create backup copies of key files that
can be used if the originals are lost. Many file systems employ stable
storage as a high-reliability strategy.

Data integrity:
Multiple users frequently share a file system. The integrity of data saved
in a shared file must be guaranteed by the file system. That is, concurrent
access requests from many users who are competing for access to the same
file must be correctly synchronized using a concurrency control method.
Atomic transactions are a high-level concurrency management mechanism
for data integrity that is frequently offered to users by a file system.

120
120
Security: Distributed System
Management
A distributed file system should be secure so that its users may trust that
their data will be kept private. To safeguard the information contained in
the file system from unwanted & unauthorized access, security
mechanisms must be implemented.

Heterogeneity:
Heterogeneity in distributed systems is unavoidable as a result of huge
scale. Users of heterogeneous distributed systems have the option of using
multiple computer platforms for different purposes.

8.1.3 History:
The server component of the Distributed File System was initially
introduced as an add-on feature. It was added to Windows NT 4.0 Server
and was known as ―DFS 4.1‖. Then later on it was included as a standard
component for all editions of Windows 2000 Server. Client-side support
has been included in Windows NT 4.0 and also in later on version of
Windows. Linux kernels 2.6.14 and versions after it come with an SMB
client VFS known as ―cifs‖ which supports DFS. Mac OS X 10.7 (lion)
and onwards supports Mac OS X DFS.

8.1.4 Applications:
NFS:
NFS stands for Network File System. It is a client-server architecture that
allows a computer user to view, store, and update files remotely. The
protocol of NFS is one of the several distributed file system standards for
Network-Attached Storage (NAS).

CIFS:
CIFS stands for Common Internet File System. CIFS is an accent of SMB.
That is, CIFS is an application of SIMB protocol, designed by Microsoft.

SMB:
SMB stands for Server Message Block. It is a protocol for sharing a file
and was invented by IMB. The SMB protocol was created to allow
computers to perform read and write operations on files to a remote host
over a Local Area Network (LAN). The directories present in the remote
host can be accessed via SMB and are called as ―shares‖.

Hadoop:
Hadoop is a group of open-source software services. It gives a software
framework for distributed storage and operating of big data using the
MapReduce programming model. The core of Hadoop contains a storage
part, known as Hadoop Distributed File System (HDFS), and an operating
part which is a MapReduce programming model.

121
Distributed System and NetWare:
Cloud Computing
NetWare is an abandon computer network operating system developed by
Novell, Inc. It primarily used combined multitasking to run different
services on a personal computer, using the IPX network protocol.

8.1.5 Working of DFS:

There are two ways in which DFS can be implemented:

Standalone DFS namespace:

It allows only for those DFS roots that exist on the local computer and are
not using Active Directory. A Standalone DFS can only be acquired on
those computers on which it is created. It does not provide any fault
liberation and cannot be linked to any other DFS. Standalone DFS roots
are rarely come across because of their limited advantage.

Domain-based DFS namespace:

It stores the configuration of DFS in Active Directory, creating the DFS
namespace root accessible
at \\<domainname>\<dfsroot> or \\<FQDN>\<dfsroot>

A file system is a set of data structures, interfaces, abstractions, and APIs

that work together to manage any type of file on any type of storage
device, in a consistent manner. Each operating system uses a particular file
system to manage the files.
In the early days, Microsoft used FAT (FAT12, FAT16, and FAT32) in
the MS-DOS and Windows 9x family. Starting from Windows NT 3.1,
Microsoft developed New Technology File System (NTFS), which had
many advantages over FAT32, such as supporting bigger files, allowing
longer filenames, data encryption, access management, journaling, and a
lot more.
122
122
NTFS has been the default file system of the Window NT family (2000, Distributed System
XP, Vista, 7, 10, etc.) ever since. NTFS isn’t suitable for non-Windows Management
environments.
For instance, you can only read the content of an NTFS-formatted
storage device (like flash memory) on a Mac OS, but you won’t be able to
write anything to it - unless you install an NTFS driver with write support.
Or you can just use the exFat file system. Extended File Allocation
Table (exFAT) is a lighter version of NTFS created by Microsoft in
2006. exFAT was designed for high-capacity removable devices, such as
external hard disks, USB drives, and memory cards. exFAT is the default
file system used by SDXC cards.
Unlike NTFS, exFAT has read and write support on Non-Windows
environments as well, including Mac OS — making it the best cross-
platform file system for high-capacity removable storage devices. So
basically, if you have a removable disk you want to use on Windows,
Mac, and Linux, you need to format it to exFAT. Apple has also
developed and used various file systems over the years, including
Hierarchical File System (HFS), HFS+, and recently Apple File
System (APFS). Just like NTFS, APFS is a journaling file system and has
been in use since the launch of OS X High Sierra in 2017. But how
about file systems in Linux distributions?
The Extended File System (ext) family of file systems was created for
the Linux kernel - the core of the Linux operating system. The first version
of ext was released in 1991, but soon after, it was replaced by the
second extended file system (ext2) in 1993. In the 2000s, the third
extended filesystem (ext3) and fourth extended filesystem
(ext4) were developed for Linux with journaling capability. ext4 is now
the default file system in many distributions of Linux, including
Debian and Ubuntu. You can use the findmnt command on Linux to list
your ext4-formatted partitions:
findmnt -lo source,target,fstype,used -t ext4
The output would be something like:
SOURCE TARGET FSTYPE USED
/dev/vda1 / ext4 3.6G

Architecture of file systems:

A file system installed on an operating system consists of three layers:

● Physical file system

● Virtual file system
● Logical file system
These layers can be implemented as independent or tightly coupled
abstractions. When people talk about file systems, they refer to one of
123
Distributed System and these layers or all three as one unit. Although these layers are different
Cloud Computing across operating systems, the concept is the same. The physical layer is the
concrete implementation of a file system; It's responsible for data storage
and retrieval and space management on the storage device (or precisely:
partitions).
The physical file system interacts with the storage hardware via device
drivers. The next layer is the virtual file system or VFS. The virtual file
system provides a consistent view of various file systems mounted on
the same operating system. So does this mean an operating system can use
multiple file systems at the same time? The answer is yes!
It's common for a removable storage medium to have a different file
system than that of a computer.
For instance, on Windows (which uses NTFS as the primary file system),
a flash memory might have been formatted to exFAT or FAT32. That said,
the operating system should provide a unified interface between
computer programs (file explorers and other apps that work with files) and
the different mounted file systems (such as NTFS, APFS, ext4, FAT32,
exFAT, and UDF).
For instance, when you open up your file explorer program, you can copy
an image from an ext4 file system and paste it over to your exFAT-
formatted flash memory - without having to know that files are managed
differently under the hood. This convenient layer between the user (you)
and the underlying file systems is provided by the VFS. A VFS defines a
contract that all physical file systems must implement to be supported by
that operating system.
However, this compliance isn't built into the file system core, meaning the
source code of a file system doesn't include support for every operating
system's VFS. Instead, it uses a file system driver to adhere to the VFS
rules of every file system. A driver is a program that enables software to
communicate with another software or hardware.
Although VFS is responsible for providing a standard interface between
programs and various file systems, computer programs don't interact with
VFS directly. Instead, they use a unified API between programs and the
VFS. Can you guess what it is? Yes, we're talking about the logical file
system.
The logical file system is the user-facing part of a file system, which
provides an API to enable user programs to perform various file
operations, such as OPEN, READ, and WRITE, without having to deal
with any storage hardware. On the other hand, VFS provides a bridge
between the logical layer (which programs interact with) and a set of the
physical layer of various file systems.

124
124
 Distributed System
Management

A high-level architecture of the file system layers

What does it mean to mount a file system?:

On Unix-like systems, the VFS assigns a device ID (for instance,
dev/disk1s1) to each partition or removable storage device. Then, it
creates a virtual directory tree and puts the content of each device
under that directory tree as separate directories. The act of assigning a
directory to a storage device (under the root directory tree) is called
mounting, and the assigned directory is called a mount point. That said,
on a Unix-like operating system, all partitions and removable storage
devices appear as if they are directories under the root directory.
For instance, on Linux, the mounting points for a removable device (such
as a memory card), are usually under the /media directory. That said, once
a flash memory is attached to the system, and consequently, auto
mounted at the default mounting point (/media in this case), its content
would be available under the /media directory. However, there are times
you need to mount a file system manually.
On Linux, it’s done like so:
mount /dev/disk1s1 /media/usb
In the above command, the first parameter is the device ID (/dev/disk1s1),
and the second parameter (/media/usb) is the mount point. Please note that
the mount point should already exist as a directory. If it doesn’t, it has to
be created first:
mkdir -p /media/usb
mount /dev/disk1s1 /media/usb
If the mount-point directory already contains files, those files will be
hidden for as long as the device is mounted.
125
Distributed System and Files metadata:
Cloud Computing
File metadata is a data structure that contains data about a file, such as:
● File size
● Timestamps, like creation date, last accessed date, and modification
date
● The file's owner
● The file's mode (who can do what with the file)
● What blocks on the partition are allocated to the file
● and a lot more
Metadata isn’t stored with the file content, though. Instead, it’s stored in a
different place on the disk - but associated with the file.
In Unix-like systems, the metadata is in the form of data structures, called
inode. Inodes are identified by a unique number called the inode number.
Inodes are associated with files in a table called inode tables. Each file on
the storage device has an inode, which contains information about it such
as the time it was created, modified, etc.The inode also includes the
address of the blocks allocated to the file; On the other hand, where
exactly it's located on the storage device In an ext4 inode, the address of
the allocated blocks is stored as a set of data structures called
extents (within the inode). Each extent contains the address of the first
data block allocated to the file and the number of the continuous
blocks that the file has occupied.
Whenever you open a file on Linux, its name is first resolved to an inode
number.
Having the inode number, the file system fetches the respective inode
from the inode table. Once the inode is fetched, the file system starts to
compose the file from the data blocks registered in the inode. You can use
the df command with the -i parameter on Linux to see the inodes (total,
used, and free) in your partitions:
df -i
The output would look like this:
udev 4116100 378 4115722 1% /dev
tmpfs 4118422 528 4117894 1% /run
/dev/vda1 6451200 175101 6276099 3% /
As you can see, the partition /dev/vda1 has a total number of 6,451,200
inodes, of which 3% have been used (175,101 inodes). To see the inodes
associated with files in a directory, you can use the ls command with -
126
126 il parameters.
ls -li Distributed System
Management
And the output would be:
1303834 -rw-r--r-- 1 root www-data 2502 Jul 8 2019 wp-links-
opml.php
1303835 -rw-r--r-- 1 root www-data 3306 Jul 8 2019 wp-load.php
1303836 -rw-r--r-- 1 root www-data 39551 Jul 8 2019 wp-login.php
1303837 -rw-r--r-- 1 root www-data 8403 Jul 8 2019 wp-mail.php
1303838 -rw-r--r-- 1 root www-data 18962 Jul 8 2019 wp-settings.php
The first column is the inode number associated with each file.
The number of inodes on a partition is decided when you format a
partition. That said, as long as you have free space and unused inodes, you
can store files on your storage device.
It's unlikely that a personal Linux OS would run out of inodes. However,
enterprise services that deal with a large number of files (like mail servers)
have to manage their inode quota smartly.
On NTFS, the metadata is stored differently, though.
NTFS keeps file information in a data structure called the Master File
Table (MFT). Every file has at least one entry in MFT, which contains
everything about it, including its location on the storage device - similar to
the inodes table. On most operating systems, you can grab metadata via
the graphical user interface. For instance, when you right-click on a file on
Mac OS, and select Get Info (Properties in Windows), a window appears
with information about the file. This information is fetched from the
respective file’s metadata.

Space Management:
Storage devices are divided into fixed-sized blocks called sectors. A
sector is the minimum storage unit on a storage device and is between
512 bytes and 4096 bytes (Advanced Format). However, file systems use a
high-level concept as the storage unit, called blocks. Blocks are an
abstraction over physical sectors; Each block usually consists of multiple
sectors. Depending on the file size, the file system allocates one or more
blocks to each file. Speaking of space management, the file system is
aware of every used and unused block on the partitions, so it’ll be able to
allocate space to new files or fetch the existing ones when requested.
The most basic storage unit in ext4-formatted partitions is the block.
However, the contiguous blocks are grouped into block groups for easier
management.

127
Distributed System and
Cloud Computing

The layout of a block group within an ext4 partition

Each block group has its own data structures and data blocks.
Here are the data structures a block group can contain:
● Super Block: a metadata repository, which contains metadata about
the entire file system, such as the total number of blocks in the file
system, total blocks in block groups, inodes, and more. Not all block
groups contain the superblock, though. A certain number of block
groups store a copy of the super as a backup.
● Group Descriptors: Group descriptors also contain bookkeeping
information for each block group
● Inode Bitmap: Each block group has its own inode quota for storing
files. A block bitmap is a data structure used to identify used and
unused inodes within the block group. 1 denotes used and
0 denotes unused inode objects.
● Block Bitmap: a data structure used to identify used & unused data
blocks within the block group. 1 denotes used and 0 denotes unused
data blocks
● Inode Table: a data structure that defines the relation of files and
their inodes. The number of inodes stored in this area is related to the
block size used by the file system.
● Data Blocks: This is the zone within the block group where file
contents are stored. Ext4 file system even takes one step further
(comparing to ext3), and organizes block groups into a bigger group
called flex block groups.
The data structures of each block group, including the block bitmap, inode
bitmap, and inode table, are concatenated and stored in the first block
group within each flex block group. Having all the data structures
concatenated in one block group (the first one) frees up more contiguous
data blocks on other block groups within each flex block group. These
concepts might be confusing, but you don't have to master every bit of
them. It's just to depict the depth of file systems. The layout of the first
block group looks like this:

128
128
 Distributed System
Management

The layout of the first block in an ext4 flex block group

When a file is being written to a disk, it is written to one or more blocks
within a block group.
Managing files at the block group level improves the performance of the
file system significantly, as opposed to organizing files as one unit.

Size vs size on disk

Have you ever noticed that your file explorer displays two different sizes
for each file: size, and size on disk.

Size and Size on disk:

Why are size and size on disk slightly different?
We already know depending on the file size, one or more blocks are
allocated to a file. One block is the minimum space that can be allocated
to a file. This means the remaining space of a partially-filled block cannot
be used by another file. This is the rule!
Since the size of the file isn’t an integer multiple of blocks, the last block
might be partially used, and the remaining space would remain unused - or
would be filled with zeros.
So "size" is basically the actual file size, while "size on disk" is the space
it has occupied, even though it’s not using it all.

129
Distributed System and You can use the du command on Linux to see it yourself.
Cloud Computing
du -b "some-file.txt"

The output would be something like this:

623 icon-link.svg
And to check the size on disk:
du -B 1 "icon-link.svg"
Which will result in:
4096 icon-link.svg
Based on the output, the allocated block is about 4kb, while the actual file
size is 623 bytes. This means each block size on this operating system is
4kb.

8.1.6 Advantages:
● DFS allows multiple user to access or store the data.
● It allows the data to be share remotely.
● It improved the availability of file, access time, and network
efficiency.
● Improved the capacity to change the size of the data and also
improves the ability to exchange the data.
● Distributed File System provides transparency of data even if server
or disk fails.

8.1.7 Disadvantages:
● In Distributed File System nodes and connections needs to be secured
therefore we can say that security is at stake.
● There is a possibility of lose of messages and data in the network
while movement from one node to another.
● Database connection in case of Distributed File System is
complicated.
● Also handling of the database is not easy in Distributed File System as
compared to a single user system.
● There are chances that overloading will take place if all nodes tries to
send data at once.

130
130
8.1.8 Benefits of DFS Models: Distributed System
Management
The distributed file system brings with it some common benefits.
A DFS makes it possible to restrict access to the file system, depending on
access lists or capabilities on both the servers and the clients, depending
on how the protocol is designed. Also, since the server also provides a
single central point of access for data requests, it is thought to be fault-
tolerant (as mentioned above) in that it will still function well if some of
the nodes are taken offline. This dovetails with some of the reasons that
DFS was developed in the first place – the system can still have that
integrity if a few workstations get moved around.

8.1.9 DFS and Backup:

Ironically enough, even though a DFS server is prized for being a single
central point of access, another server may also be in play. However, that
doesn't mean that there won't be that single central access point. The
second server will be for backup.
Because businesses invest in having one central DFS server, they will
worry that the server could be compromised somehow. Backing all of the
data up at a separate location ensures the right kind of redundancy to make
the system fully fault-tolerant, even if the king itself (the primary server)
is toppled by something like a DDoS attack or something else. DFS
systems, like other systems, continue to innovate. With new kinds of net

8.1.10 The challenges associated with DFS:

● Data redundancy and inconsistency.
● Difficulty in accessing data.
● Data isolation
● Integrity problems
● Unauthorized access is not restricted.
● It coordinates only physical access.

8.1.11 Components:
The components of DFS are as follows:
● Block Storage provider
● Client Driver
● Security provider
● Meta- Data Service
● Object service.

131
Distributed System and These components are pictorially represented below:
Cloud Computing

Features:
The features of DFS are as follows:
● User mobility
● Easy to use
● High availability
● Performance
● Coherent access
● Location independence
● File locking
● Multi-networking access
● Local gateways
● Multi-protocol access

Example:
Given below is an example of DFS Structure:

132
132
 Distributed System
Management

8.2 FILE MODELS IN DISTRIBUTED SYSTEM

Several machines are utilized in Distributed File Systems (DFS) to
supply the file system's facility. Various file systems frequently use
various conceptual models. Models based on structure and mobility is
frequently used for file modeling. In this article, you will learn about the
file models in the distributed operating system.
Types of Files models in the distributed operating systems

133
Distributed System and There are mainly two types of file models in the distributed operating
Cloud Computing system.

1. Structure Criteria
2. Modifiability Criteria
Structure Criteria
There are two types of file models in structure criteria. These are as
follows:

1. Structured Files
2. Unstructured Files
Structured Files
The Structured file model is presently a rarely used file model. In the
structured file model, a file is seen as a collection of records by the file
system. Files come in various shapes and sizes and with a variety of
features. It is also possible that records from various files in the same file
system have varying sizes. Despite belonging to the same file system, files
have various attributes. A record is the smallest unit of data from which
data may be accessed. The read/write actions are executed on a set of
records. Different "File Attributes" are provided in a hierarchical file
system to characterize the file. Each attribute consists of two parts: a name
and a value. The file system used determines the file attributes. It provides
information on files, file sizes, file owners, the date of last modification,
the date of file creation, access permission, and the date of last access.
Because of the varied access rights, the Directory Service function is
utilized to manage file attributes.

The structured files are also divided into two types:

1. Files with Non-Indexed records
2. Files with Indexed records

1. Files with Non-Indexed records:

Records in non-indexed files are retrieved based on their placement inside
the file. For instance, the second record from the starting and the second
from the end of the record.

2. Files with Indexed records:

Each record contains a single or many key fields in a file containing
indexed records, each of which may be accessed by specifying its value. A
file is stored as a B-tree or similar data structure or hash table to find
records quickly.

134
134
Unstructured Files: Distributed System
Management
It is the most important and widely used file model. A file is a group of
unstructured data sequences in the unstructured model. Any substructure
does not support it. The data and structure of each file available in the file
system is an uninterrupted sequence of bytes such as UNIX or DOS. Most
latest OS prefer the unstructured file model instead of the structured file
model due to sharing of files by multiple apps. It has no structure;
therefore, it can be interpreted in various ways by different applications.

Modifiability Criteria:
There are two files model in the Modifiability Criteria. These are as
follows:
1. Mutable Files
2. Immutable Files

1. Mutable Files:
The existing operating system employs the mutable file model. A file is
described as a single series of records because the same file is updated
repeatedly once new material is added. After a file is updated, the existing
contents are changed by the new contents.

2. Immutable Files:
The Immutable file model is used by Cedar File System (CFS). The file
may not be modified once created in the immutable file model. Only after
the file has been created can it be deleted. Several versions of the same file
are created to implement file updates. When a file is changed, a new file
version is created. There is consistent sharing because only immutable
files are shared in this file paradigm. Distributed systems allow caching
and replication strategies, overcoming the limitation of many copies and
maintaining consistency. The disadvantages of employing the immutable
file model include increased space use and disc allocation activity. CFS
uses the "Keep" parameter to keep track of the file's current version
number. When the parameter value is 1, it results in the production of a
new file version. The previous version is erased, and the disk space is
reused for a new one. When the parameter value is greater than 1, it
indicates the existence of several versions of a file. If the version number
is not specified, CFS utilizes the lowest version number for actions such
as "delete" and the highest version number for other activities such
as "open".
The specific client's request for accessing a particular file is serviced on
the basis of the file accessing model used by the distributed file system.
The file accessing model basically depends on 1 ) the unit of data access
and 2 ) the method used for accessing remote files.
On the basis of the unit of data access, following file access models might
be used in order to access the specific file.
135
Distributed System and 1. File-level transfer model
Cloud Computing
2. Block-level transfer model
3. Byte-level transfer model
4. Record-level transfer model
1. File-level transfer model: In file-level transfer model, the complete
file is moved while a particular operation necessitates the file data to
be transmitted all the way through the distributed computing network
amongst client and server. This model has better scalability and is
efficient.
2. Block-level transfer model: In block-level transfer model, file data
transfers through the network amongst client and a server is
accomplished in units of file blocks. In short, the unit of data transfer
in block-level transfer model is file blocks. The block-level transfer
model might be used in distributed computing environment
comprising several diskless workstations.
3. Byte-level transfer model: In byte-level transfer model, file data
transfers the network amongst client and a server is accomplished in
units of bytes. In short, the unit of data transfer in byte-level transfer
model is bytes. The byte-level transfer model offers more flexibility
in comparison to the other file transfer models since, it allows
retrieval and storage of an arbitrary sequential subrange of a file. The
major disadvantage of byte-level transfer model is the trouble in cache
management because of the variable-length data for different access
requests.
4. Record-level transfer model: The record-level file transfer model
might be used in the file models where the file contents are structured
in the form of records. In record-level transfer model, file data
transfers through the network amongst client and a server is
accomplished in units of records. The unit of data transfer in record-
level transfer model is record.

8.3 SUMMARY
DFS allows multiple user to access or store the data. It allows the data
to be share remotely. It improved the availability of file, access time, and
network efficiency. Improved the capacity to change the size of the data
and also improves the ability to exchange the data.
The main purpose of the Distributed File System (DFS) is to allows users
of physically distributed systems to share their data and resources by using
a Common File System.
A collection of workstations and mainframes connected by a Local Area
Network (LAN) is a configuration on Distributed File System. A DFS is
executed as a part of the operating system. In DFS, a namespace is created
136
136 and this process is transparent for the clients.
8.4 REFERENCE FOR FURTHER READING Distributed System
Management
1. https://www.unf.edu/~sahuja/cis6302/filesystems.html
2. https://www.geeksforgeeks.org/what-is-dfsdistributed-file-system/
3. https://www.ques10.com/p/2247/what-are-the-good-features-of-a-
distributed-file-1/
4. https://www.javatpoint.com/distributed-file-system
5. https://en.wikipedia.org/wiki/Distributed_File_System_(Microsoft)
6. https://erandipraboda.medium.com/characteristics-of-distributed-file-
systems-bf5988f85d3

8.5 MODEL QUESTIONS

1. What is DFS (Distributed File System)?
2. What are the Features of DFS
3. Discuss on the History of DFS?
4. What are the Applications of DFS?
5. Explain the Working of DFS?
6. What are the Advantages of DFS?
7. What are the Disadvantages of DFS?
8. Explain the Benefits of DFS Models?
9. Discuss on DFS and Backup?
10. Discuss The challenges associated with DFS?
11. What are the Components of DFS?
12. Explain the concept of File Models in Distributed System?

*****

137
 UNIT V

9
INTRODUCTION TO CLOUD
COMPUTING
Unit Structure
9.0 Objective
9.1 Introduction
9.1.1 History and evolution
9.2 Characteristics of cloud computing
9.3 Cloud Computing example
9.4 Benefits of Cloud Computing
9.5 Risks of Cloud Computing
9.6 Cloud Computing Architecture
9.6.1 Cloud Architecture model
9.6.2 Types of Cloud
9.6.3 Cloud based Services
9.6.3.1 Software as a service (SaaS)
9.6.3.2 Platform as a service (PaaS)
9.6.3.3 Infrastructure as a service (IaaS)
9.7 Summary
9.8 Referances

9.0 OBJECTIVE
After studying this module, you will be able to understand
● Cloud Computing and its characteristics
● Benefits of Cloud Computing
● Cloud Computing Architecture
● Cloud based Services

9.1 INTRODUCTION
● Cloud Computing is a model for enabling convenient, on-demand
network access to a shared pool of resources that can be rapidly
provided and released with minimum management efforts.
● Cloud Computing intends to realize the concept of Computing as a
utility like water, gas, electricity etc.

138
● Cloud Computing referred as the accessing and storing of data and Introduction to Cloud
provide services related to computing over the internet. Computing

● It simply referred as it remote services on the internet manage and

access data online rather than any local drives.
● The data can be anything like images, videos, audios, documents, files
etc.
● The potential of cloud computing has been recognized by industry.
● Cloud computing is in huge demand so, big organization providing
the service like Amazon AWS, Microsoft Azure, Google
Cloud, Alibaba cloud etc. are some Cloud Computing service Provide

9.1.1 History and evolution:

● In early days before Cloud Computing was come into existence, client
Server Architecture was used.
● In this all the data and control of client resides in Server side, so if
user wants to access data it has to take the permission. It is having
many disadvantages.
● So, After Client Server computing, Distributed Computing was come
into existence.
● In this all the computers are in network so user can share resources as
and when required. This also having certain limitations.
● So to remove limitations faced in distributed system, cloud computing
was emerged.
● During 1961, John MacCharty delivered his speech at MIT that
“Computing Can be sold as a Utility.
● This concept of computing was not appreciated so after few years this
is implemented by Salesforce.com in 1999.
● This company started delivering an enterprise application over the
internet and in this way Cloud Computing was started.
● In 2002, Amazon started Amazon Web Services (AWS), Amazon will
provide storage, computation over the internet.
● In 2006 Amazon will launch Elastic Compute Cloud Commercial
Service which is open for Everybody to use.
● In 2009, Google Play also started providing Cloud Computing
Enterprise Application.
● In 2009, Microsoft launch Microsoft Azure and after that other
companies like Alibaba, IBM, Oracle, HP also introduces their Cloud
Services.

139
Distributed System and
Cloud Computing
9.2 CHARACTERISTICS OF CLOUD COMPUTING
1. On demand self-service Here the user is able to use web services
and resources on demand. User can
logon to the website any time and use
them.
2. Ubiquitous Access- As Cloud Computing is completely web
based user can have accessed it from
anywhere and anytime.
3. Resource Pooling- Resource pooling allows cloud
providers to pool large-scale IT
resources to serve multiple cloud
consumers. Different physical and
virtual IT resources are dynamically
assigned and reassigned according to
cloud consumer demand, typically
followed by execution through statistical
multiple xing.
4. Rapid Elasticity Elasticity is the automated ability of
acloud to transparent scale IT
resources, as required in response to
runtime conditions or as pre-determined
by the cloud consumer or cloud provider.
5. Measured Usage The measured usage characteristic
represents the ability of a Cloud
platform to keep track of the usage of its
IT resources, primarily by cloud
consumers. Based on what is measured,
the cloud provider can charge a cloud
consumer only for the IT resources used
and/or for the time frame during which
access to the IT resources was granted.
6. Resiliency Resilient computing is a form of failover
that distributes redundant
Implementations of IT resources across
physical locations.

9.3 CLOUD COMPUTING EXAMPLES

Here we are discussing about two popular cloud computing facilities:

1. Amazon Elastic Computing Cloud (EC2):

 It is a part of set of standalone services which includes object storage
service, for hosting and simple database.

140
140
 With EC2 user may rent virtual machine instances to run their own Introduction to Cloud
software and also can monitor number of VMs as demand changes. Computing

 If user wants to use Amazon EC2 then he has to create Amazon

Machine Image(AMI), then upload AMI to Amazon S3, select OS and
start instances, then monitor and control via web interface or API.

2. Google App Engine:

 It is an end –to-end service.

 It allows developer to run their web application on Googles

infrastructure for that

 Download App engine SDK

 Develop an application as a set of python programs

 Register for an application ID

 Submit the application to Google

9.4 BENEFITS OF CLOUD COMPUTING

Various advantages are provided by Cloud Computing

1. On-demand access to pay-as-you-go computing resources on a short-

term basis (such as processors by the hour), and the ability to release
these computing resources when they are no longer needed.

2. The perception of having unlimited computing resources that are

available on demand, Thereby reducing the need to prepare for
provisioning.

3. The ability to add or remove IT resources at a fine-grained level,

such as modifying Available storage disk space by single gigabyte
increments.

4. Abstraction of the infrastructure so applications are not locked into

devices or location And can be easily moved if needed

5. An IT resource with increased availability is accessible for longer

periods of time (for example, 22 hours out of a 24-hour day).
Cloud providers generally offer “resilient” IT resources for which
they are able to guarantee high levels of availability.

6. An IT resource with increased reliability is able to better avoid &

recover from exception conditions. The modular architecture of cloud
environments provides extensive fail over support that increases
reliability.

141
Distributed System and
Cloud Computing

9.5 RISKS OF CLOUD COMPUTING

Some downsides of Cloud Computing are discussed below:
1. Security and privacy:
It is always a risk to handover the sensitive information since data
management and infrastructure management in cloud computing by third
party.

2. Lock in:
To switch from one Cloud Service provider(CSP) to another is very
difficult. It results in dependency on particular CSP.

3. Isolation Failure:
It involves the failure of isolation mechanism which will separate storage,
memory, routing between different tenants.

4. Management Interface compromise:

In case of public cloud provider, customer management interface is
accessible through the internet.

5. Insecure data deletion:

Sometime data requested for deletion may not get deleted. This can
happen because extra copies of data are stored but not available on disk or
disk may destroyed.

9.6 CLOUD COMPUTING ARCHITECTURE

In current time Cloud Computing is giving a new shape to every
organization by providing on demand virtualized services/resources.
Starting from small to medium and medium to large, every organization
use cloud computing services in storing information and accessing that
from anywhere and anytime only with the help of internet. Transparency,
scalability, security and intelligent monitoring are some of the most
important constraints which every cloud infrastructure should experience.

9.6.1 Cloud Computing Architecture:

Cloud Computing Architecture:
The cloud architecture is divided into 2 parts i.e.
1. Frontend
2. Backend
142
142
 Introduction to Cloud
Computing

Figure - Internal Architectural view of Cloud Computing

1. Frontend:
Frontend of the cloud architecture refers to the client side of cloud
computing system. Means it contains all the user interfaces and
applications which are used by the client to access the cloud computing
services/resources.
For example, use of a web browser to access the cloud plat form.

Client Infrastructure:
Client Infrastructure refers to the frontend components. It contains the
applications and user interfaces which are required to access the cloud
platform.

2. Backend:
Backend refers to the cloud itself which is used by the service provider. It
contains the resources as well as manages the resources and provides
security mechanisms. Along with this it includes huge storage, virtual
applications, virtual machines, traffic control mechanisms, deployment
models etc.

1. Application:
Application in backend refers to a software or platform to which client
accesses. Means it provides the service in backend as per the client
requirement.

2. Service:
Service in backend refers to the major three types of cloud based services
like SaaS, PaaS and IaaS. Also manages which type of service the user
accesses.

143
Distributed System and 3. Cloud Runtime:
Cloud Computing
Runtime cloud in backend refers to provide of execution and runtime
platform/environment to the virtual machine.

4. Storage:
Storage in backend refers to provide flexible and scalable storage service
and management of stored data.

5. Infrastructure:
Cloud Infrastructure in backend refers to hardware and software
components of cloud like it includes servers, storage, network devices,
virtualization software etc.

6. Management:
Management in backend refers to management of backend components
like application, service, runtime cloud, storage, infrastructure, and other
security mechanisms etc.

7. Security:
Security in backend refers to implementation of different security
mechanisms in the backend for secure cloud resources, systems, files, and
infrastructure to end-users.

8. Internet:
Internet connection acts as the medium or a bridge between frontend and
backend and establishes the interaction and communication between
frontend and backend.

Benefits of Cloud Computing Architecture:

● Makes overall cloud computing system simpler.
● Improves data processing requirements.
● Helps in providing high security.
● Makes it more modularized.
● Results better disaster recovery.
● Gives good user accessibility.
● Reduces IT operating costs.

9.6.2 Types of Cloud:

A cloud deployment model represents a specific type of cloud
environment, primarily distinguished by ownership, size, and access.
There are four common cloud deployment models:
144
144
• Public cloud Introduction to Cloud
Computing
• Community cloud
• Private cloud
• Hybrid cloud

Public Cloud:
 A public cloud is a publicly accessible cloud environment owned by a
third-party cloud provider
 The IT resources on public clouds are usually provisioned via cloud
delivery models and are generally offered to cloud consumers at a
cost or are commercialized via other avenues (such as
advertisement).
 The cloud provider is responsible for the creation and on-going
maintenance of the public cloud and its IT resources.
 Example-Google, Oracle, Microsoft

Community Clouds:
 A community cloud is similar to a public cloud except that its
access is limited to a specific community of cloud consumers.
 The community cloud may be jointly owned by the community
members or by third-party cloud provider that provisions a public
cloud with limited access.
 The member cloud consumers of the community typically share the
responsibility for defining and evolving the community Cloud.
 Example- Government agency
Private Clouds:
 A private cloud is owned by a single organization.
 Private clouds enable an organization to use cloud computing
technology as a means of centralizing access to IT resources by
different parts, locations, or departments the organization.
 The use of a private cloud can change how organizational and trust
boundaries defined and applied
 The actual administration of a private cloud environment may be
carried out by internal or outsourced staff.
 Example-HP data centre, Ubuntu

145
Distributed System and Hybrid Clouds:
Cloud Computing
 A hybrid cloud is a cloud environment comprised of two or more
different cloud deployment models.
 The service of a hybrid cloud can be distributed in multiple cloud
types.
 Example-Amazon Web service

9.6.3 Cloud based Services:

A cloud delivery model represent a specific, pre-packaged combination of
IT resources offered by a cloud provider. Three common cloud delivery
models have become widely established and formalized:
• Software-as-a-Service (SaaS)
• Platform-as-a-Service (PaaS)
• Infrastructure-as-a-Service (IaaS)

9.6.3.1 Software as a service (SaaS):

 A software program positioned as a shared cloud service and made
available as a “product” or generic utility represents the typical profile
of a SaaS offering.
 The SaaS delivery model is typicallyused to make a reusable cloud
service widely available (often commercially) to a range of cloud
consumers.
 SaaS application includes billing and invoice system, CRM, Help
desk application, HR solutions.
 SaaS provides Application programming Interface(API), which
allows the developer to develop a customized application.

Figure 1: The cloud service consumer is given access the cloud

service contract, but not to any underlying IT resources or
implementation details

146
146
Benefits: Introduction to Cloud
Computing
1. It is beneficial in terms of scalability, efficiency, performance.
2. Modest software tools
3. Efficient use of software licence
4. Centralized management and data
5. Platform responsibility managed by provider
6. Multitenant solution

9.6.3.2 Platform as a service (PaaS):

 The PaaS delivery model represents a pre-defined “ready-to-use”
environment typically comprised of already deployed and configured
IT resources.
 PaaS relies on (and is primarily defined by) the usage of a readymade
environment that establishes a set of pre-packaged products and tools
used to support the entire delivery lifecycle of custom applications.
 Common reasons a cloud consumer would use and invest in a PaaS
environment include:
 The cloud consumer wants to extend on premise environments in
to the cloud for scalability and economic purposes.
 The cloud consumer uses the ready-made environment to entirely
substitute an on premise environment.
 The cloud consumer wants to become a cloud provider and deploys
its own cloud services to be made available to other external cloud
consumers.

147
Distributed System and Figure 2: A cloud consumer is accessing a ready-made PaaS
Cloud Computing environment. The question mark indicates that the cloud consumer is
intentionally shielded from the implementation details of the
platform.
Benefits:
1. Lower administrative overhead:
Consumer need not to bother much about the administrative because its
responsibility of cloud provider.

2. Lower total cost of ownership:

Consumer need not purchase expensive hardware, servers, power and data
storage.

3. Scalable Solution:
It is easy to scale up and down automatically based upon application
resources

Issues:
1. Lack of portability between PaaS clouds
2. Event based processor scheduling
3. Security engineering of PaaS application

9.6.3.3 Infrastructure as a service (IaaS)

 The IaaS delivery model represents a self-contained IT environment
comprised of infrastructure-centric IT resources that can be accessed
and managed via cloud service-based interfaces and tools.
 This environment can include hardware, network, connectivity,
operating systems And other “raw” IT resources.
 The general purpose of an IaaS environment is to provide cloud
consumer with high level of control and responsibility over its
configuration and utilization.
 The IT resources provided by IaaS are generally not pre-configured,
placing administrative responsibility directly upon the cloud
consumer
 This model is therefore used by cloud consumers that require a high
level of control over the cloud-based environment they intend to
create.

148
148
 Introduction to Cloud
Computing

Figure 3: A cloud consumer is using a virtual server within an

IaaS environment. Cloud consumers are provided with a range of
contractual guarantees by The cloud provider, pertaining to
characteristics such as capacity, performance, and availability.

Benefits:
Some of the key benefits of IaaS are listed below:
1. Full control of computing resources through administrative access
to VMs:
 It allows the consumer to access computing resources through
administrative access to virtual machine in the following manner
 Consumer issues administrative command to cloud provide to run the
virtual machine or to save data on cloud server.
▪ Consumer issues administrative command to virtual machines they
owned to start web or installing new application.

2. Flexible and efficient renting of computer hardware

▪ IaaS resources such as virtual machine, storage, bandwidth, IP
address, monitoring services, firewall all are made available to the
consumer on rent.
3. Portability, interoperability with legacy application
▪ It is possible to maintain legacy between application and workloads
between IaaS cloud.

Issues:
1. Compatibility with legacy security vulnerability
2. Virtual machine Sprawl
3. Robustness of VM level isolation
4. Data erase practices

Comparing Cloud Delivery Models:

A comparison of typical cloud delivery model control levels

149
Distributed System and
Cloud Computing

Typical activities carried out by cloud consumers and cloud providers in

relation to the cloud delivery models.

9.7 SUMMARY
 Cloud environments are comprised of highly extensive infrastructure
that Offers pools of IT resources that can be leased using a pay-for-
use model whereby Only the actual usage of the IT resources is
billable.

 Cloud environments can introduce distinct security challenges, some

of which pertain to overlapping trust boundaries imposed by a cloud
provider sharing IT resources with multiple cloud consumers.

 A cloud consumer’s operational governance can be limited within

cloud environments due to the control exercised by a cloud provider
over its platforms.

 The portability of cloud-based IT resources can be inhibited by

dependencies upon proprietary characteristics imposed by a
cloud.

 The geographical location of data and IT resources can be out of a

cloud consumer’s control when hosted by a third-party cloud
150
150
 provider. This can introduce various legal and regulatory Introduction to Cloud
compliance concerns. Computing

 The IaaS cloud delivery model offers cloud consumers a high level of
administrative control over “raw” infrastructure-based IT resources.

 The PaaS cloud delivery model enables a cloud provider to offer a

preconfigured environment that cloud consumers can use to build and
deploy cloud services and solutions, albeit with decreased
administrative control.

 SaaS is a cloud delivery model for shared cloud services that can be
positioned as commercialized products hosted by clouds.

 Different combinations of IaaS, PaaS, and SaaS are possible,

depending on how cloud consumers and cloud providers choose to
leverage the natural hierarchy established by these base cloud delivery
models.

 A public cloud is owned by a third party and generally offers

commercialized cloud services and IT resources to cloud consumer
organizations.

 A private cloud is owned by an individual organization and resides

within the organization’s premises.

 A community cloud is normally limited for access by a group of cloud

consumers that may also share responsibility in its ownership.

 A hybrid cloud is a combination of two or more other cloud

deployment models.
Self-Learning Topics: Cluster computing, Grid computing, Fog
computing

9.8 REFERENCES-
 Cloud Computing Concepts, Technology & Architecture by Thomas
Erl, Zaigham Mahmood, and Ricardo Puttini

 James Broberg and Andrzej M. Goscinski, Cloud Computing:

Principles and Paradigms Wiley, First edition, ISBN No. 978-04-708-
8799-8

 Rajkumar Buyya, Christian Vecchiola, S. ThamaraiSelvi, Mastering

Cloud Computing, Tata Mcgraw Hill, ISBN No. 978-12-590-2995-0

*****

151
 UNIT VI

10
CLOUD COMPUTING
Unit Structure
10.0 Objective
10.1 Introduction
10.2 Amazon Web Services
10.3 Microsoft Azure and Google Cloud
10.3.1 Compute services
10.3.2 Storage services
10.3.3 Database services
10.3.4 Additional services

10.0 OBJECTIVE
The main motivation behind cloud computing is to enable businesses to
get access to data centres and manage tasks from a remote location. Cloud
computing works on the pay-as-you-go pricing model, which helps
businesses lower their operating cost and run infrastructure more
efficiently.

What is cloud computing?:

Cloud computing is a general term for anything that involves delivering
hosted services over the internet. These services are divided into three
main categories or types of cloud computing: infrastructure as a service
(IaaS), platform as a service (PaaS) and software as a service (SaaS).
A cloud can be private or public. A public cloud sells services to anyone
on the internet. A private cloud is a proprietary network or a data center
that supplies hosted services to a limited number of people, with certain
access and permissions settings. Private or public, the goal of cloud
computing is to provide easy, scalable access to computing resources and
IT services.
Cloud infrastructure involves the hardware and software components
required for proper implementation of a cloud computing model. Cloud
computing can also be thought of as utility computing or on-demand
computing.
The name cloud computing was inspired by the cloud symbol that's often
used to represent the internet in flowcharts and diagrams.

152
How does cloud computing work?: Cloud Computing

Cloud computing works by enabling client devices to access data and

cloud applications over the internet from remote physical servers,
databases and computers.

10.1 INTRODUCTION
Cloud Computing is the delivery of computing services such as servers,
storage, databases, networking, software, analytics, intelligence, and more,
over the Cloud (Internet).

Cloud Computing provides an alternative to the on-premises datacentre.

With an on-premises datacentre, we have to manage everything, such as
purchasing and installing hardware, virtualization, installing the operating
system, and any other required applications, setting up the network,
configuring the firewall, and setting up storage for data. After doing all the
set-up, we become responsible for maintaining it through its entire
lifecycle.
But if we choose Cloud Computing, a cloud vendor is responsible for the
hardware purchase and maintenance. They also provide a wide variety of
software and platform as a service. We can take any required services on
rent. The cloud computing services will be charged based on usage.

153
Distributed System and
Cloud Computing

The cloud environment provides an easily accessible online portal that

makes handy for the user to manage the compute, storage, network, and
application resources. Some cloud service providers are in the following
figure.

Advantages of cloud computing:

 Cost: It reduces the huge capital costs of buying hardware and
software.
 Speed: Resources can be accessed in minutes, typically within a few
clicks.

154
154
 Scalability: We can increase or decrease the requirement of resources Cloud Computing
according to the business requirements.
 Productivity: While using cloud computing, we put less operational
effort. We do not need to apply patching, as well as no need to
maintain hardware and software. So, in this way, the IT team can be
more productive and focus on achieving business goals.
 Reliability: Backup and recovery of data are less expensive and very
fast for business continuity.
 Security: Many cloud vendors offer a broad set of policies,
technologies, and controls that strengthen our data security.

Types of Cloud Computing:

 Public Cloud: The cloud resources that are owned and operated by a
third-party cloud service provider are termed as public clouds. It
delivers computing resources such as servers, software, and storage
over the internet
 Private Cloud: The cloud computing resources that are exclusively
used inside a single business or organization are termed as a private
cloud. A private cloud may physically be located on the company’s
on-site datacentre or hosted by a third-party service provider.
 Hybrid Cloud: It is the combination of public and private clouds,
which is bounded together by technology that allows data applications
to be shared between them. Hybrid cloud provides flexibility and
more deployment options to the business.

Types of Cloud Services:

155
Distributed System and
Cloud Computing

1. Infrastructure as a Service (IaaS):

In IaaS, we can rent IT infrastructures like servers and virtual machines
(VMs), storage, networks, operating systems from a cloud service vendor.
We can create VM running Windows or Linux and install anything we
want on it. Using IaaS, we don’t need to care about the hardware or
virtualization software, but other than that, we do have to manage
everything else. Using IaaS, we get maximum flexibility, but still, we need
to put more effort into maintenance.

2. Platform as a Service (PaaS):

This service provides an on-demand environment for developing, testing,
delivering, and managing software applications. The developer is
responsible for the application, and the PaaS vendor provides the ability to
deploy and run it. Using PaaS, the flexibility gets reduce, but the
management of the environment is taken care of by the cloud vendors.

3. Software as a Service (SaaS):

It provides a centrally hosted and managed software services to the end-
users. It delivers software over the internet, on-demand, and typically on a
subscription basis. E.g., Microsoft One Drive, Dropbox, WordPress,
Office 365, and Amazon Kindle. SaaS is used to minimize the operational
cost to the maximum extent.

156
156
 Cloud Computing

10.2 AMAZON WEB SERVICES

AWS stands for Amazon Web Services which uses distributed IT
infrastructure to provide different IT resources on demand.
Our AWS tutorial includes all the topics such as introduction, history of
aws, global infrastructure, features of AWS, IAM, Storage services,
Database services, etc.

What are AWS?:

o AWS stands for Amazon Web Services.
o The AWS service is provided by the Amazon that uses distributed IT
infrastructure to provide different IT resources available on demand.
It provides different services such as infrastructure as a service (IaaS),
platform as a service (PaaS) and packaged software as a service
(SaaS).
o Amazon launched AWS, a cloud computing platform to allow the
different organizations to take advantage of reliable IT infrastructure.

Uses of AWS:
o A small manufacturing organization uses their expertise to expand
their business by leaving their IT management to the AWS.
o A large enterprise spread across the globe can utilize the AWS to
deliver the training to the distributed workforce.
o An architecture consulting company can use AWS to get the high-
compute rendering of construction prototype.

157
Distributed System and o A media company can use the AWS to provide different types of
Cloud Computing content such as ebox or audio files to the worldwide files.

Pay-As-You-Go:
Based on the concept of Pay-As-You-Go, AWS provides the services to
the customers.
AWS provides services to customers when required without any prior
commitment or upfront investment. Pay-As-You-Go enables the
customers to procure services from AWS.
o Computing
o Programming models
o Database storage
o Networking

Advantages of AWS:
1) Flexibility:
o We can get more time for core business tasks due to the instant
availability of new features and services in AWS.
o It provides effortless hosting of legacy applications. AWS does not
require learning new technologies and migration of applications to the
AWS provides the advanced computing and efficient storage.
o AWS also offers a choice that whether we want to run the applications
and services together or not. We can also choose to run a part of the
IT infrastructure in AWS and the remaining part in data centres.

2) Cost-effectiveness:
AWS requires no upfront investment, long-term commitment, and
minimum expense when compared to traditional IT infrastructure that
requires a huge investment.

3) Scalability/Elasticity:
Through AWS, autoscaling and elastic load balancing techniques are
automatically scaled up or down, when demand increases or decreases
respectively. AWS techniques are ideal for handling unpredictable or very
high loads. Due to this reason, organizations enjoy the benefits of reduced
cost and increased user satisfaction.

4) Security:
 AWS provides end-to-end security and privacy to customers.
 AWS has a virtual infrastructure that offers optimum availability
while managing full privacy and isolation of their operations.
158
158
 Customers can expect high-level of physical security because of Cloud Computing
Amazon's several years of experience in designing, developing and
maintaining large-scale IT operation centers.
 AWS ensures the three aspects of security, i.e., Confidentiality,
integrity, and availability of user's data.

10.3 MICROSOFT AZURE AND GOOGLE CLOUD

What is Azure:
Microsoft Azure is a growing set of cloud computing services created by
Microsoft that hosts your existing applications, streamline the
development of a new application, and also enhances our on-premises
applications. It helps the organizations in building, testing, deploying, and
managing applications and services through Microsoft-managed data
centers.

Azure Services:

 Compute services: It includes the Microsoft Azure Cloud Services,

Azure Virtual Machines, Azure Website, and Azure Mobile Services,
which processes the data on the cloud with the help of powerful
processors.

 Data services: This service is used to store data over the cloud that
can be scaled according to the requirements. It includes Microsoft
Azure Storage (Blob, Queue Table, and Azure File services), Azure
SQL Database, and the Redis Cache.

 Application services: It includes services, which help us to build and

operate our application, like the Azure Active Directory, Service Bus
for connecting distributed systems, HDInsight for processing big data,
the Azure Scheduler, and the Azure Media Services.

 Network services: It helps you to connect with the cloud and on-
premises infrastructure, which includes Virtual Networks, Azure
Content Delivery Network, and the Azure Traffic Manager.

Why Should You Choose Microsoft Azure Services?:

While every business can have their individual reasons to choose
Microsoft Azure services, there are several unbeatable advantages that all
can leverage upon, like:
● Microsoft Azure is capable of providing an enticing combination of
Infrastructure as a platform (IaaS) and Platform as a service (PaaS)
that helps enterprises create and deploy their own web apps without
hustling over the groundwork.
● Microsoft Azure has designed Security Development Lifecycle, an
industry-standard security process that considers all security features

159
Distributed System and including getting licenses and ensuring the best safety in all
Cloud Computing operations.
● There is a vast user base already on Microsoft Azure, but the
infrastructure is constantly scaling up, by using more processes for
applications and selling storage through the cloud. It can run without
any additional coding.
● A hybrid cloud computing ecosystem is still a unique feature of
Microsoft Azure. It can improve the performance by utilizing Virtual
Private Networks (VPNs), ExpressRoute connections, caches, CDNs,
etc.
● As most enterprises rely on MS office tools, it is wise to invest in a
cloud platform that integrates well with all Microsoft products.
Additionally, knowing C++, C, and Visual Basic can help you steer
your career in Microsoft Azure. If you require further validation, then
you can try out the Azure certification courses for Windows
certificates.
● Microsoft Azure has intelligence and analytics capacities to improve
the business process with the help of machine learning bots, cognitive
APIs, and Blockchain as a Service (BaaS).
● Microsoft Azure also has SQL and noSQL data processing facilities to
get deeper and actionable insights from the available data.
● One of the major reasons to choose Azure services is the affordability,
as the virtual infrastructure maintenance is extremely cost-efficient.

Curated List of Top Azure Services:

While there is no long list of competitors in cloud servicing, the top
runners like Google and AWS continue to give a tough fight to Microsoft
Azure in the race of being the most used cloud service. Despite intense
competition, Microsoft Azure continues growing and evolving over the
years, especially through the phase of remote working due to a pandemic
in 2020 and 2021. Offering top Azure services, the platform has
maintained its integrity and popularity. Now let’s delve deeper into
understanding more about the top 10 most popular Azure services.

How Azure works:

It is essential to understand the internal workings of Azure so that we can
design our applications on Azure effectively with high availability, data
residency, resilience, etc.

160
160
 Cloud Computing

Microsoft Azure is completely based on the concept of virtualization. So,

similar to other virtualized data center, it also contains racks. Each rack
has a separate power unit and network switch, and also each rack is
integrated with a software called Fabric-Controller. This Fabric-
controller is a distributed application, which is responsible for managing
and monitoring servers within the rack. In case of any server failure, the
Fabric-controller recognizes it and recovers it. And Each of these Fabric-
Controller is, in turn, connected to a piece of software called Orchestrator.
This Orchestrator includes web-services, Rest API to create, update, and
delete resources.
When a request is made by the user either using PowerShell or Azure
portal. First, it will go to the Orchestrator, where it will fundamentally do
three things:
1. Authenticate the User
2. It will Authorize the user, i.e., it will check whether the user is
allowed to do the requested task.
3. It will look into the database for the availability of space based on the
resources and pass the request to an appropriate Azure Fabric
controller to execute the request.
Combinations of racks form a cluster. We have multiple clusters within a
data center, and we can have multiple Data Centers within an Availability
zone, multiple Availability zones within a Region, and multiple Regions
within a Geography.

 Geographies: It is a discrete market, typically contains two or more

regions, that preserves data residency and compliance boundaries.

 Azure regions: A region is a collection of data centers deployed

within a defined perimeter and interconnected through a dedicated
regional low-latency network.

161
Distributed System and Azure covers more global regions than any other cloud provider, which
Cloud Computing offers the scalability needed to bring applications and users closer around
the world. It is globally available in 50 regions around the world. Due to
its availability over many regions, it helps in preserving data residency and
offers comprehensive compliance and flexible options to the customers.

10.3.1 Compute services:

Azure compute services are the hosting services responsible for hosting
and running the application workloads. These include Azure Virtual
Machines (VMs), Azure Container Service, Azure App Services, Azure
Batch, and Azure ServiceFabric.
Azure Compute Services: Azure Virtual Machines (VMs) and Azure
Container Service:

Azure Virtual Machines (VMs):

A Microsoft Azure Virtual Machine (VM) is an on-demand, scalable
computing resource. You don’t need to buy any physical hardware and
bear its maintenance cost; you have the flexibility of virtualization. Your
cloud administrators only need to select the operating system, configure
the required resources, and create the web server – all this gets done
within a few minutes.

Azure Container Service:

Azure helps you leverage the modern container-based development
practices and microservices architecture. You can migrate your .NET
applications to microservices using Windows Server containers with
Azure Service Fabric. Further, you can use Azure Kubernetes Service to
scale and orchestrate Linux Containers.
You can choose between Docker Hub and Azure Container Registry to
store your images and deploy to any preferred target. Moreover, it
simplifies the configuration process and optimizes it for the cloud. The
major advantage is that it consumes less space as compared to VMs and
starts instantly; hence speeding up the processes.

10.3.2 Storage services:

The Azure Storage platform is Microsoft's cloud storage solution for
modern data storage scenarios. Azure Storage offers highly available,
massively scalable, durable, and secure storage for a variety of data
objects in the cloud. Azure Storage data objects are accessible from
anywhere in the world over HTTP or HTTPS via a REST API. Azure
Storage also offers client libraries for developers building applications or
services with .NET, Java, Python, JavaScript, C++, and Go. Developers
and IT professionals can use Azure PowerShell and Azure CLI to write
scripts for data management or configuration tasks. The Azure portal and
Azure Storage Explorer provide user-interface tools for interacting with
Azure Storage.
162
162
Benefits of Azure Storage: Cloud Computing

Azure Storage services offer the following benefits for application

developers and IT professionals:
● Durable and highly available: Redundancy ensures that your data is
safe in the event of transient hardware failures. You can also opt to
replicate data across data centers or geographical regions for
additional protection from local catastrophe or natural disaster. Data
replicated in this way remains highly available in the event of an
unexpected outage.
● Secure: All data written to an Azure storage account is encrypted by
the service. Azure Storage provides you with fine-grained control
over who has access to your data.
● Scalable: Azure Storage is designed to be massively scalable to meet
the data storage and performance needs of today's applications.
● Managed: Azure handles hardware maintenance, updates, and critical
issues for you.
● Accessible: Data in Azure Storage is accessible from anywhere in the
world over HTTP or HTTPS. Microsoft provides client libraries for
Azure Storage in a variety of languages, including .NET, Java,
Node.js, Python, PHP, Ruby, Go, and others, as well as a mature
REST API. Azure Storage supports scripting in Azure PowerShell or
Azure CLI. And the Azure portal and Azure Storage Explorer offer
easy visual solutions for working with your data.

Azure Storage data services:

The Azure Storage platform includes the following data services:
● Azure Blobs: A massively scalable object store for text and binary
data. Also includes support for big data analytics through Data Lake
Storage Gen2.
● Azure Files: Managed file shares for cloud or on-premises
deployments.
● Azure Queues: A messaging store for reliable messaging between
application components.
● Azure Tables: A NoSQL store for schema-less storage of structured
data.
● Azure Disks: Block-level storage volumes for Azure VMs.

10.3.3 Database services:

163
Distributed System and The basic fundamental building block that is available in Azure is the SQL
Cloud Computing database. Microsoft offers this SQL server and SQL database on Azure in
many ways. We can deploy a single database, or we can deploy multiple
databases as part of a shared elastic pool.

Azure Database Service Architecture:

Microsoft introduced a managed instance that is targeted towards on-
premises customers. So, if we have some SQL databases within our on-
premises datacentre and we want to migrate that database into Azure
without any complex configuration, or ambiguity, then we can use
managed instance. Because this is mainly targeted towards on-premises
customers who want to lift and share their on-premises database into
Azure with the least effort and optimized cost. We can also take advantage
of licensing we have within our on-premises data center.

Microsoft will be responsible for maintenance patching and related

services. But, in case if we want to go for the IaaS service for the SQL
server, then we can deploy SQL Server on the Azure Virtual machine. If
the data have a dependency on the underlying platform and we want to log
into the SQL Server, in that case, we can use the SQL server on a virtual
machine.
We can deploy a SQL data warehouse on the cloud. Azure offers many
other database services for different types of databases such as MySQL,
Maria DB, and also PostgreSQL. Once we deployed a database into
Azure, we need to migrate the data into it or replicate the data into it.

Azure Database Services for Data Migration:

The services that are available in Azure, which we can use to migrate the
data from our on-premises SQL Server into Azure.
Azure Data Migration Service: It is used to migrate the data from our
existing SQL server and database within the on-premises data center into
Azure.
164
164
Azure SQL data sync: If we want to replicate the data from our on- Cloud Computing
premises database into Azure, then we can use Azure SQL data sync.
SQL Stretch Database: It is used to migrate cold data into Azure. SQL
stretch database is a bit different from other database offerings. It works as
a hybrid database because it divides the data into different types - hot and
cold. A hot data will be kept in the on-premises data center and cold data
in the Azure.

Data Factory:
It is used for ETL transformation, extraction loading, etc. Using the data
factory, we can even extract the data from our on-premises data center.
We can do some conversion and load it into the Azure SQL database. Data
Factory is an ETL tool that is offered on the cloud, which we can use to
connect to different databases, extract the data, transform it, and load into
a destination.

Security:
All the databases that are existing in Azure need to be secured, and also
we need to accept connections from known origins. For this purpose, all
these database services come with firewall rules where we can configure
from which particular IP address we want to allow connections. We can
define those firewall rules to limit the number of connections and also
reduce the surface attack area.

Cosmos DB:
Cosmos DB is a NoSQL data store that is available in Azure, and it is
designed to be globally scalable and also very highly available with
extremely low latency. Microsoft guarantees latency in terms of reading
and writes with Cosmos DB. For example - if we have any applications
such as IoT, gaming where we get a lot of data from different users spread
across globally, then we will go for Cosmos DB. Because Cosmos DB is
designed to be globally scalable and highly available due to which our
users will experience low latency.
Finally, there are two things, and one is we need to secure all the services.
For that purpose, we can integrate all these services with Azure Active
Directory and manage the users from Azure Active Directory also. To
monitor all these services, we can use the security center. There is an
individual monitoring tool too, but Azure security center will keep on
monitoring all these services and provide recommendations if something
is wrong.

Cosmos DB:
Cosmos DB is a NoSQL data store that is available in Azure, and it is
designed to be globally scalable and also very highly available with
extremely low latency. Microsoft guarantees latency in terms of reading
and writes with Cosmos DB. For example - if we have any applications
such as IoT, gaming where we get a lot of data from different users spread
165
Distributed System and across globally, then we will go for Cosmos DB. Because Cosmos DB is
Cloud Computing designed to be globally scalable and highly available due to which our
users will experience low latency.
Finally, there are two things, and one is we need to secure all the services.
For that purpose, we can integrate all these services with Azure Active
Directory and manage the users from Azure Active Directory also. To
monitor all these services, we can use the security center. There is an
individual monitoring tool too, but Azure security center will keep on
monitoring all these services and provide recommendations if something
is wrong.

10.3.4 Additional services:

Azure data services store and manage data on cloud. Microsoft Azure
comes with a range of data services: Azure Storage, Azure SQL Database,
Azure Document DB, Azure StorSimple, and Azure Redis Cache.
Below is an overview of some most popular Microsoft Azure services,
specifically the top 10 azure services and the way you can use them across
the entire architecture:
● Azure DevOps
● Azure Blob Storage
● Azure Virtual Machines
● Azure Backup
● Azure Cosmos DB
● Azure Logic Apps
● Azure Active Directory
● API management
● Azure Content Delivery Network
● Azure Site Recovery
● Azure Bots

*****

166
166
 11
CLOUD PLATFORMS
Unit Structure
11.0 Objective
11.1 Introduction
11.2 Google App Engine (GAE)
11.3 Aneka
11.4 Comparative study of various Cloud
11.5 Computing Platforms

11.0 OBJECTIVE
A computer platform is a system that consists of a hardware device and
an operating system that an application, program or process runs upon. An
example of a computer platform is a desktop computer with Microsoft
Windows installed on it. A desktop is a hardware device and Windows is
an operating system.
The operating system acts as an interface between the computer and the
user and also between the computer and the application. So, in order to
have a functional device, you need hardware and an operating system
together to make a usable computer platform for a program to run on.
The hardware portion of a computer platform consists of a processor,
memory, and storage. The processor is a bit like your brain and memory is
like a scratchpad for your brain to use while you're working out a problem.
It used to be that people referred to different computer platforms by their
physical size, from smallest to largest - microcomputers (smallest),
minicomputers (mid-size), and mainframes (largest). The term
microcomputer has fallen somewhat out of favor - now most people just
refer to these machines as computers or personal computers.

11.1 INTRODUCTION
Cloud computing is the delivery of computing services—including
servers, storage, databases, networking, software, analytics, and
intelligence—over the Internet ("the cloud") to offer faster innovation,
flexible resources, and economies of scale.

How does cloud computing work?:

Rather than owning their own computing infrastructure or data centres,
companies can rent access to anything from applications to storage from a
cloud service provider.

167
Distributed System and One benefit of using cloud-computing services is that firms can avoid the
Cloud Computing upfront cost and complexity of owning and maintaining their own IT
infrastructure, and instead simply pay for what they use, when they use it.
In turn, providers of cloud-computing services can benefit from significant
economies of scale by delivering the same services to a wide range of
customers.

What cloud-computing services are available?:

Cloud-computing services cover a vast range of options now, from the
basics of storage, networking and processing power, through to natural
language processing and artificial intelligence as well as standard office
applications. Pretty much any service that doesn't require you to be
physically close to the computer hardware that you are using can now be
delivered via the cloud – even quantum computing.

What are examples of cloud computing?:

Cloud computing underpins a vast number of services. That includes
consumer services like Gmail or the cloud backup of the photos on your
smartphone, though to the services that allow large enterprises to host all
their data and run all of their applications in the cloud. For example,
Netflix relies on cloud-computing services to run its its video-streaming
service and its other business systems, too.
Cloud computing is becoming the default option for many apps: software
vendors are increasingly offering their applications as services over the
internet rather than standalone products as they try to switch to a
subscription model. However, there are potential downsides to cloud
computing, in that it can also introduce new costs and new risks for
companies using it.

Why is it called cloud computing?:

A fundamental concept behind cloud computing is that the location of the
service, and many of the details such as the hardware or operating system
on which it is running, are largely irrelevant to the user. It's with this in
mind that the metaphor of the cloud was borrowed from old telecoms
network schematics, in which the public telephone network (and later the
internet) was often represented as a cloud to denote that the location didn't
matter – it was just a cloud of stuff. This is an over-simplification of
course; for many customers, location of their services and data remains a
key issue.

What is the history of cloud computing?:

Cloud computing as a term has been around since the early 2000s, but the
concept of computing as a service has been around for much, much longer
– as far back as the 1960s, when computer bureaus would allow
companies to rent time on a mainframe, rather than have to buy one
themselves.
168
168
These 'time-sharing' services were largely overtaken by the rise of the PC, Cloud Platforms
which made owning a computer much more affordable, and then in turn
by the rise of corporate data centres where companies would store vast
amounts of data.
But the concept of renting access to computing power has resurfaced again
and again – in the application service providers, utility computing, and
grid computing of the late 1990s and early 2000s. This was followed by
cloud computing, which really took hold with the emergence of software
as a service and hyperscale cloud-computing providers such as Amazon
Web Services.

How important is the cloud?:

Building the infrastructure to support cloud computing now accounts for a
significant chunk of all IT spending, while spending on traditional, in-
house IT slides as computing workloads continue to move to the cloud,
whether that is public cloud services offered by vendors or private clouds
built by enterprises themselves.
Indeed, it's increasingly clear that when it comes to enterprise computing
platforms, like it or not, the cloud has won.
Tech analyst Gartner predicts that as much as half of spending across
application software, infrastructure software, business process services
and system infrastructure markets will have shifted to the cloud by 2025,
up from 41% in 2022. It estimates that almost two-thirds of spending on
application software will be via cloud computing, up from 57.7% in 2022.

11.2 GOOGLE APP ENGINE (GAE)

Google App Engine lets you run (host) your own Web applications on
Google’s infrastructure. However, by no means is this a ―rent a piece of a
server‖ hosting service. With App Engine, your application is not hosted
on a single server. There are no servers to maintain: You just upload your
application, and it’s ready to serve your users. Just as servicing a Google
search request may involve dozens, or even hundreds of Google servers, 169
Distributed System and all totally hidden and satisfied in a fraction of a second, Google App
Cloud Computing Engine applications run the same way, on the same infrastructure. This is
the unique aspect of Google’s approach. Yes, you cede some control to
Google, but you are rewarded by being totally free of the infrastructure,
capacity management, and load balancing tasks that enterprise typically
have to manage, irrespective of whether they are self-hosting or hosting on
someone else’s PaaS or IaaS.
You can choose to share your application with the world, or limit access to
members of your organization. Google App Engine supports apps written
in several programming languages:
With App Engine’s Java runtime environment, you can build your app
using standard Java technologies, including the JVM, Java servlets, and
the Java programming language—or any other language using a JVM-
based interpreter or compiler, such as JavaScript or Ruby. App Engine
also features a dedicated Python runtime environment, which includes a
fast Python interpreter and the Python standard library. The Java and
Python runtime environments are built to ensure that your application runs
quickly, securely, and without interference from other apps on the system.
As with most cloud-hosting services, with App Engine, you only pay for
what you use. Google levies no set-up costs and no recurring fees. Similar
to Amazon’s AWS, resources such as storage and bandwidth are measured
by the gigabyte.
App Engine costs nothing to get started. All applications can use up to 500
MB of storage and enough CPU and bandwidth to support an efficient app
serving around 5 million page views a month, absolutely free. When you
enable billing for your application, your free limits are raised, and you
only pay for resources you use above the free levels.
Application developers have access to persistent storage technologies such
as the Google File System (GFS) and Bigtable, a distributed storage
system for unstructured data. The Java version supports asynchronous
nonblocking queries using the Twig Object Datastore interface. This offers
an alternative to using threads for parallel data processing.
―With Google App Engine, developers can write Web applications based
on the same building blocks that Google uses,‖ Kevin Gibbs, Google’s
technical lead for the project, wrote in The Official Google Blog―Google.
Twig is an object persistence interface built on Google App Engine’s low-
level datastore which overcomes many of JDO-GAEs limitations,
including full support for inheritance, polymorphism, and generic types.
You can easily configure, modify or extend Twig’s behavior by
implementing your own strategies or overriding extension points in pure
Java code. App Engine packages those building blocks and provides
access to scalable infrastructure that we hope will make it easier for
developers to scale their applications automatically as they grow.‖
Google App Engine has appeared at a time when an increasing number of
tech companies are moving their operations to the cloud; it places Google
170
170
squarely in competition with Amazon’s Elastic Cloud Computing (EC2) Cloud Platforms
and Simple Storage Service (S3) offerings.
Google says its vision with Google App Engine is to offer developers a
more holistic, end-to-end solution for building and scaling applications
online. Its servers are configured to balance the load of traffic to
developers’ applications, scaling to meet the demand of an influx of
traffic. App Engine also includes APIs for user authentication to allow
developers to sign on for services, and for e-mail, to manage
Google Cloud Platform provides a wide range of computing services that
target broad categories of user needs.
The Google Cloud Platform provides mainly 6 types of compute options:
1. App Engine
2. Compute Engine
3. Kubernetes Engine
4. Cloud Functions
5. Cloud Run
6. VMware Engine

171
Distributed System and Now let’s talk about some of these services in brief.
Cloud Computing
Compute Engine:
The Compute Engine service is Google’s unmanaged compute service. We
can think of Compute Engine as an Infrastructure as a Service (IaaS)
offering by Google Cloud. As the service is unmanaged, it is our
responsibility to configure, administer, and monitor the system. On
Google’s side, they will ensure that resources are available, reliable, and
ready for you to use. The main benefit in using compute engine is that you
have complete control of the systems.

You can do the following when you build on Compute Engine:

● Create Virtual Instances, which is the smallest unit in the GCP
project.
● Create instance groups to easily manage multiple instances together.
● Create virtual machine images.
The virtual machine instances running in zones assigned to them. Zones
are data center-like resources. They are located within regions which is a
geographical location. The zones are within a region are linked with low-
latency and high bandwidth network connections.
Pros of Compute Engine:
● It offers the users complete control over the Virtual Machine
instances.
● It is easy to set up, you can spin up a server within few minutes.
● The use of preemptive VM’s can reduce the cost by up to 80%.
● Set of predefined VM configurations and VM images are available
ready to be used according to needs.
Cons of Compute Engine:
● Requires high expertise level, since everything needs to be installed
and configured by yourself.
● Autoscaling is slower than App Engine.
● To enable monitoring, you need to install packages into the VM
instances. No direct Stackdriver monitoring is possible.

App Engine:
The App Engine is Google’s Platform as a Service(PaaS) offering. It is a
compute service that provides a managed platform for running
applications. As this is a managed service, your focus should be on the
application only and Google will manage the resources needed to run the
application. Thus App Engine users have less to manage, but you will
have less control over the compute resources. The applications hosted on
App Engine are highly scalable and run reliably even under heavy load.
172
172
The App Engine supports the following languages: Cloud Platforms

● Python
● Go
● Ruby
● PHP
● Node.js
● Java
● .NET
The App Engine provides two types of runtime environments: standard
and flexible.
1. The Standard environment provides a secured and sandboxed
environment for running applications and distributes requests across
multiple servers to meet the demand. The applications run
independently of the hardware, OS, and physical location of the
server.
2. The Flexible environment provides more options and control to the
developers who want to use App Engine, but without the language
constraints of the standard environment. It uses Docker containers as
the basic building blocks. These containers can be auto-scaled
according to load.

Pros of App Engine:

● You need to focus only on the application code, the rest of everything
is managed by Google. Thus reducing management complexities.
● As it provides version management, thus it is easy to maintain and roll
out versions of applications.
● It has faster autoscaling as the size of instances is smaller.
● Easy to deploy and monitor.

Cons of App Engine:

● It is more constrained as the instances are smaller, thus enabling fast
autoscaling, but there can be cases when large applications require
larger instances.
● As it is a fully managed service, the user has no control over the
underlying infrastructure that may be required for some complex
applications.
● It is expensive in the long run as the cost adds up quickly.

173
Distributed System and Difference between Compute Engine and App Engine:
Cloud Computing
Compute Engine App Engine
Service model IaaS offering PaaS offering
Type of Service Unmanaged Service Managed Service
Control over More control and Less control over
resources flexibility computing resources
Costs Costs less than App Costs more than
Engine over resources Compute Engine
Running Instances When running Can scale down to zero
application, at least one instances when no
instance should be requests are coming
running
Use cases Best for general Best for web-facing
computing workloads and mobile
applications
Autoscaling Slower autoscaling Faster autoscaling
Security Less secure than App Comparatively more
Engine secure than Compute
Engine

11.3 ANEKA
Aneka includes an extensible set of APIs associated with programming
models like MapReduce.
These APIs support different cloud models like a private, public, hybrid
Cloud.
Manjrasoft focuses on creating innovative software technologies to
simplify the development and deployment of private or public cloud
applications. Our product plays the role of an application platform as a
service for multiple cloud computing.

Multiple Structures:
Aneka is a software platform for developing cloud computing
applications.
In Aneka, cloud applications are executed.
Aneka is a pure PaaS solution for cloud computing.
Aneka is a cloud middleware product.
Manya can be deployed over a network of computers, a multicore server, a
data center, a virtual cloud infrastructure, or a combination thereof.

174
174
Multiple containers can be classified into three major categories: Cloud Platforms

1. Textile services
2. Foundation Services
3.Application Services

1. Textile Services:

Fabric Services defines the lowest level of the software stack that
represents multiple containers. They provide access to resource-
provisioning subsystems and monitoring features implemented in many.

2. Foundation Services:

Fabric Services are the core services of Manya Cloud and define the
infrastructure management features of the system. Foundation services are
concerned with the logical management of a distributed system built on
top of the infrastructure and provide ancillary services for delivering
applications.

3. Application Services:

Application services manage the execution of applications and constitute a

layer that varies according to the specific programming model used to
develop distributed applications on top of Aneka.

There are mainly two major components in multiple technologies:

The SDK (Software Development Kit) includes the Application

Programming Interface (API) and tools needed for the rapid development
of applications. The Anka API supports three popular cloud programming
models: Tasks, Threads and MapReduce;

And:

175
Distributed System and A runtime engine and platform for managing the deployment and
Cloud Computing execution of applications on a private or public cloud.

One of the notable features of Aneka Pass is to support the provision of

private cloud resources from desktop, cluster to a virtual data center
using VMware, Citrix Zen Server, and public cloud resources such
as Windows Azure, Amazon EC2, and GoGrid cloud service.

Aneka's potential as a Platform as a Service has been successfully

harnessed by its users and customers in three different areas,
including engineering, life sciences, education, and business
intelligence.

Architecture of Aneka:

Aneka is a platform and framework for developing distributed applications

on the Cloud. It uses desktop PCs on-demand and CPU cycles in addition
to a heterogeneous network of servers or datacenters. Aneka provides a
rich set of APIs for developers to transparently exploit such resources and
express the business logic of applications using preferred programming
abstractions.

System administrators can leverage a collection of tools to monitor and

control the deployed infrastructure. It can be a public cloud available to
anyone via the Internet or a private cloud formed by nodes with restricted
access.

176
176
A multiplex-based computing cloud is a collection of physical and Cloud Platforms
virtualized resources connected via a network, either the Internet or a
private intranet. Each resource hosts an instance of multiple containers
that represent the runtime environment where distributed applications are
executed. The container provides the basic management features of a
single node and takes advantage of all the other functions of its hosting
services.

Services are divided into clothing, foundation, and execution services.

Foundation services identify the core system of Aneka middleware, which
provides a set of infrastructure features to enable Anka containers to
perform specific and specific tasks. Fabric services interact directly with
nodes through the Platform Abstraction Layer (PAL) and perform
hardware profiling and dynamic resource provisioning. Execution services
deal directly with scheduling and executing applications in the Cloud.

One of the key features of Aneka is its ability to provide a variety of ways
to express distributed applications by offering different programming
models; Execution services are mostly concerned with providing
middleware with the implementation of these models. Additional services
such as persistence and security are inverse to the whole stack of services
hosted by the container.

At the application level, a set of different components and tools are

provided to

 Simplify the development of applications (SDKs),

 Port existing applications to the Cloud, and

 Monitor and manage multiple clouds.

An Aneka-based cloud is formed by interconnected resources that are

dynamically modified according to user needs using resource
virtualization or additional CPU cycles for desktop machines. A common
deployment of Aneka is presented on the side. If the deployment identifies
a private cloud, all resources are in-house, for example, within the
enterprise.

This deployment is enhanced by connecting publicly available on-demand

resources or by interacting with several other public clouds that provide
computing resources connected over the Internet.

177
Distributed System and 11.4 COMPARATIVE STUDY OF VARIOUS CLOUD
Cloud Computing
Below table shows the comparitive study of various cloud:

11.5 COMPUTING PLATFORMS

What Does Platform Mean?:
A platform is a group of technologies that are used as a base upon which
other applications, processes or technologies are developed.
In personal computing, a platform is the basic hardware (computer) and
software (operating system) on which software applications can be run.
This environment constitutes the basic foundation upon which any
application or software is supported and/or developed.
Computers use specific central processing units (CPUs) that are designed
to run specific machine language code. In order for the computer to run
software applications, the applications must be in that CPU’s binary-coded
machine language.
Thus, historically, application programs written for one platform would
not work on a different platform.
A computer platform — also called digital platform or computing platform
— generally refers to the operating system and computer hardware only.
An example of a computing platform is a modern laptop running Windows
as an operating system. Another example would be an Apple computer
running the Mac OS X operating system.
178
178
Platform Standards: Cloud Platforms
The platform conforms to a set of standards that enable software
developers to develop software applications for the platform. These same
standards allow owners and managers to purchase appropriate applications
and hardware. Thus, to run a bookkeeping program on a computer, one
must purchase a bookkeeping software application that was developed for
the platform on which it will be used.
Multiple Platforms:
New standards-based interfaces and open interfaces allow application
programs to run on multiple platforms. Additionally, software developers
have developed software tools that allow applications to run on multiple
platforms.
Cross-Platform Software and Multi-Platform Software:
This has given rise to the terms cross-platform software and multi-
platform software. A classic example is represented by videogames
developed specifically for a certain platform, in this case a console such as
the PlayStation or Xbox. Although the same game may exist in different
versions to be run on different systems, if that version is built to be run on
Microsoft Windows, it won’t work if loaded on an Xbox. Each gaming
platform will adhere to its own set of standards as well as rules and
hardware restrictions. For example, developers may need to lower in-game
graphics settings if the game engine is too heavy on a specific console’s
hardware.
Browsers:
Newer web browsers allow third-party plug-ins to be run as part of the
browser. Therefore, some browsers are now spoken of as platforms since
they are used as a base on which to run other applications' software
programs.
Mobile Platforms:
Today, new mobile devices such as smartphones and tablets possess their
own software and hardware. They operate independently of other systems
and are capable of running their own apps, tools, and other software,
hence they can be effectively considered as platforms.
Digital Platforms:
Software stacks and some applications are also sometimes referred as
digital platforms.
For example, SQL is a database application that is often used as an
environment to run other tools for CRM, analytics and log management.
Similarly, the collection of the three-open source applications
Elasticsearch, Logstash, and Kibana constitutes the ELK Stack, a platform
used for logging purposes.
*****

179
 12
CLOUD ISSUES AND CHALLENGES
Unit Structure
12.0 Objective
12.1 Introduction
12.2 Cloud computing issues and challenges
12.2.1 Security
12.2.2 Elasticity
12.2.3 Resource management and scheduling
12.3 Quality of service (QoS) and Resource allocation
12.4 Identity and Access management

12.0 OBJECTIVE
 To study more about cloud issues and challenges.
 To study about cloud security and elasticity.
 To study and understand about Quality of Service (QoS) and resource
allocation.

12.1 INTRODUCTION
In Simplest terms, cloud computing means storing and accessing the data
and programs on remote servers that are hosted on internet instead of
computer’s hard drive or local server. Cloud computing is also referred as
Internet based computing.

Cloud Computing Architecture:

Cloud computing architecture refers to the components and sub
components required for cloud computing. These component typically
refer to:
1. Front end(fat client, thin client)
2. Back end platforms(servers, storage)
3. Cloud based delivery and a network (Internet, Intranet, Inter cloud).

Hosting a cloud:
There are three layers in cloud computing. Companies use these layers
based on the service they provide.
● Infrastructure
● Platform
● Application
 Cloud Issues and Challenges

Three layers of Cloud Computing:

At the bottom is the foundation, the Infrastructure where the people start
and begin to build. This is the layer where the cloud hosting lives.

Now, let’s have a look at hosting:

Let’s say you have a company and a website and the website has a lot of
communications that are exchanged between members. You start with a
few members talking with each other and then gradually the numbers of
members increase.
As the time passes, as the number of members increases, there would be
more traffic on the network and your server will get slow down. This
would cause a problem.
A few years ago, the websites are put in the server somewhere, in this way
you have to run around or buy and set number of servers. It costs a lot of
money and takes lot of time. You pay for these servers when you are using
and as well as when you are not using. This is called hosting.
This problem is overcome by cloud hosting. With Cloud Computing, you
have access to computing power when you needed. Now, your website is
put in the cloud server as you put it on dedicated server. People start
visiting your website and if you suddenly need more computing power,
you would scale up according to the need.

Benefits of Cloud Hosting:

1. Scalability: With Cloud hosting, it is easy to grow and shrink the
number and size of servers based on the need.
This is done by either increasing or decreasing the resources in the
cloud. This ability to alter plans due to fluctuation in business size and
needs is a superb benefit of cloud computing especially when
experiencing a sudden growth in demand.
2. Instant: Whatever you want is instantly available in the cloud.
3. Save Money: An advantage of cloud computing is the reduction in
hardware cost. Instead of purchasing in-house equipment, hardware 181
Distributed System and needs are left to the vendor. For companies that are growing rapidly,
Cloud Computing new hardware can be a large, expensive, and inconvenience. Cloud
computing alleviates these issues because resources can be acquired
quickly and easily. Even better, the cost of repairing or replacing
equipment is passed to the vendors.
Along with purchase cost, off-site hardware cuts internal power costs
and saves space. Large data centers can take up precious office space
and produce a large amount of heat. Moving to cloud applications or
storage can help maximize space and significantly cut energy
expenditures.
4. Reliability: Rather than being hosted on one single instances of a
physical server, hosting is delivered on a virtual partition which draws
its resource, such as disk space, from an extensive network of
underlying physical servers. If one server goes offline it will have no
effect on availability, as the virtual servers will continue to pull
resource from the remaining network of servers.
5. Physical Security: The underlying physical servers are still housed
within data centres and so benefit from the security measures that
those facilities implement to prevent people accessing or disrupting
them on-site

12.2 CLOUD COMPUTING ISSUES AND CHALLENGES

Let us dive into the challenges of cloud computing, With a multitude of
benefits of implementing cloud computing in all-size businesses, cloud
computing has become a popular trend in the market. To put in simple
words: Cloud computing is nothing but moving on-premises computing to
the internet. It saves both time and money. People around the globe get
access to an open pool of resources like apps, services, servers, data, and
computer networks. It is made possible either by using a privately-owned
cloud or a 3rd-party server. It improves the way data is accessed and
removes inconsistency in further updates. Also, a minimal amount of
administration is required. Cloud computing also ensures data security,
better data storage, increased synchronization between employees, and
flexibility. Organizations have become capable of making better decisions
to scale and grow.
Despite all the development and potential of cloud computing services,
there are multiple challenges of cloud computing services that businesses
face. Here we have compiled a list of challenges of cloud computing that
need to be taken care of, to leverage the maximum capability of the cloud.
Let us get started:
1. Security
2. Password Security
3. Cost Management
182
182
4. Lack of expertise Cloud Issues and Challenges

5. Internet Connectivity
6. Control or Governance
7. Compliance
8. Multiple Cloud Management
9. Creating a private cloud
10. Performance
11. Migration
12. Interoperability and Portability
13. Reliability and High Availability
14. Hybrid-Cloud Complexity

1. SECURITY:
The topmost concern in investing in cloud services is security issues in
cloud computing. It is because your data gets stored and processed by a
third-party vendor and you cannot see it. Every day or the other, you get
informed about broken authentication, compromised credentials, account
hacking, data breaches, etc. in a particular organization. It makes you a
little more skeptical.
Fortunately, the cloud providers, these days have started to put efforts to
improve security capabilities. You can be cautious as well by verifying if
the provider implements a safe user identity management system and
access control procedures. Also, ensure it implements database security
and privacy protocols.
2. PASSWORD SECURITY:
As large numbers of people access your cloud account, it becomes
vulnerable. Anybody who knows your password or hacks into your cloud
will be able to access your confidential information.
Here the organization should use a multiple level authentication and
ensure that the passwords remain protected. Also, the passwords should be
modified regularly, especially when a particular employee resigns and
leave the organization. Access rights to usernames and passwords should
be given judiciously.

3. COST MANAGEMENT:
Cloud computing enables you to access application software over a fast
internet connection and lets you save on investing in costly computer
hardware, software, management and maintenance. This makes it
affordable. But what is challenging and expensive is tuning the
183
Distributed System and organization’s needs on the third-party platform. Another costly affair is
Cloud Computing the cost of transferring data to a public cloud, especially for a small
business or project.

4. LACK OF EXPERTISE:
With the increasing workload on cloud technologies and continuously
improving cloud tools, the management has become difficult. There has
been a consistent demand for a trained workforce who can deal with cloud
computing tools and services. Hence, firms need to train their IT staff to
minimize this challenge.

5. INTERNET CONNECTIVITY:
The cloud services are dependent on a high-speed internet connection. So
the businesses that are relatively small and face connectivity issues should
ideally first invest in a good internet connection so that no downtime
happens. It is because internet downtime might incur vast business losses.

6. CONTROL OR GOVERNANCE:
Another ethical issue in cloud computing is maintaining proper control
over asset management and maintenance. There should be a dedicated
team to ensure that the assets used to implement cloud services are used
according to agreed policies and dedicated procedures. There should be
proper maintenance and that the assets are used to meet your
organization’s goals successfully.

7. COMPLIANCE:
Another major risk of cloud computing is maintaining compliance. By
compliance we mean, a set of rules about what data is allowed to be
moved and what should be kept in-house to maintain compliance. The
organizations must follow and respect the compliance rules set by various
government bodies.

8. MULTIPLE CLOUD MANAGEMENT:

Companies have started to invest in multiple public clouds, multiple
private clouds or a combination of both called the hybrid cloud. This has
grown rapidly in recent times. So it has become important to list
challenges faced by such organizations and find solutions to grow with the
trend.

9. CREATING A PRIVATE CLOUD:

Implementing an internal cloud is advantageous. This is because all the
data remains secure in-house. But the challenge here is that the IT team
has to build and fix everything by themselves. Also, the team needs to
ensure smooth functioning of the cloud. They need to automate maximum
manual tasks. The execution of tasks should be in the correct order.
So, at the moment, it sounds quite difficult to set up a private cloud all by
184
184 yourself. But many organizations are planning to do so in future.
10. PERFORMANCE: Cloud Issues and Challenges

When your business applications move to a cloud or a third-party vendor,

so your business performance starts to depend on your provider as well.
Another major problem in cloud computing is investing in the right cloud
service provider. Before investment, you should look for providers with
innovatory technologies. The performance of the BI’s and other cloud-
based systems are linked to the provider’s systems as well. Be cautious
about choosing the provider and investigate that they have protocols to
mitigate issues that arise in real-time.

11. MIGRATION:
Migration is nothing but moving a new application or an existing
application to a cloud. In the case of a new application, the process is
pretty straightforward. But if it is an age-old company application, it
becomes tedious.

12. INTEROPERABILITY AND PORTABILITY:

Another challenge of cloud computing is that applications need to be
easily migrated between cloud providers without being locked for a set
period. There is a lack of flexibility in moving from one cloud provider to
another because of the complexity involved. Changing cloud inventions
bring a slew of new challenges like managing data movement and
establishing a secure network from scratch. Another challenge is that
customers can’t access it from everywhere, but this can be fixed by the
cloud provider so that the customer can securely access the cloud from
anywhere.

13. RELIABILITY AND HIGH AVAILABILITY:

Some of the most pressing issues in cloud computing is the need for high
availability (HA) and reliability. Reliability refers to the likelihood that a
system will be up and running at any given point in time, whereas
availability refers to how likely it is that the system will be up and running
at any given point in time. Because most businesses are now reliant on
third-party services, cloud systems must be dependable and robust. Cloud
providers continue to lack round-the-clock service, resulting in frequent
outages. It is critical to use internal or third-party tools to monitor the
service being provided. It is critical to have plans in place to monitor
SLAs, usage, robustness, performance, and business reliance on these
services.

14. HYBRID-CLOUD COMPLEXITY:

For any company, a hybrid cloud environment is often a messy mix of
multiple cloud application development and cloud service providers, as
well as private and public clouds, all operating at once. A common user
interface, consistent data, and analytical benefits for businesses are all
missing from these complex cloud ecosystems. Cloud computing

185
Distributed System and challenges such as scalability, integration, and disaster recovery are
Cloud Computing magnified in a hybrid cloud environment.

12.2.1 Security:
Cloud Computing is a type of technology that provides remote services on
the internet to manage, access, and store data rather than storing it on
Servers or local drives. This technology is also known as Serverless
technology. Here the data can be anything like Image, Audio, video,
documents, files, etc.

Need of Cloud Computing:

Before using Cloud Computing, most of the large as well as small IT
companies use traditional methods i.e. they store data in Server, and they
need a separate Server room for that. In that Server Room, there should be
a database server, mail server, firewalls, routers, modems, high net speed
devices, etc. For that IT companies have to spend lots of money. In order
to reduce all the problems with cost Cloud computing come into existence
and most companies shift to this technology.

Security Issues in Cloud Computing:

There is no doubt that Cloud Computing provides various Advantages but
there are also some security issues in cloud computing. Below are some
following Security Issues in Cloud Computing as follows.

186
186
Data Loss: Cloud Issues and Challenges

Data Loss is one of the issues faced in Cloud Computing. This is also
known as Data Leakage. As we know that our sensitive data is in the
hands of Somebody else, and we don’t have full control over our database.
So if the security of cloud service is to break by hackers then it may be
possible that hackers will get access to our sensitive data or personal files.

Interference of Hackers and Insecure API’s:

As we know if we are talking about the cloud and its services it means we
are talking about the Internet. Also, we know that the easiest way to
communicate with Cloud is using API. So it is important to protect the
Interface’s and API’s which are used by an external user. But also in cloud
computing, few services are available in the public domain. An is the
vulnerable part of Cloud Computing because it may be possible that these
services are accessed by some third parties. So it may be possible that with
the help of these services hackers can easily hack or harm our data.

User Account Hijacking:

Account Hijacking is the most serious security issue in Cloud Computing.
If somehow the Account of User or an Organization is hijacked by
Hacker. Then the hacker has full authority to perform Unauthorized
Activities.

Changing Service Provider:

Vendor lock In is also an important Security issue in Cloud Computing.
Many organizations will face different problems while shifting from one
vendor to another. For example, An Organization wants to shift
from AWS Cloud to Google Cloud Services then they ace various
problem’s like shifting of all data, also both cloud services have different
techniques and functions, so they also face problems regarding that. Also,
it may be possible that the charges of AWS are different from Google
Cloud, etc.

Lack of Skill:
While working, shifting o another service provider, need an extra feature,
how to use a feature, etc. are the main problems caused in IT Company
who doesn’t have skilled Employee. So it requires a skilled person to work
with cloud Computing.

Denial of Service (DoS) attack:

This type of attack occurs when the system receives too much traffic.
Mostly DoS attacks occur in large organizations such as the banking
sector, government sector, etc. When a DoS attack occurs data is lost. So,
in order to recover data, it requires a great amount of money as well as
time to handle it.

187
Distributed System and 12.2.2 Elasticity:
Cloud Computing
The Elasticity refers to the ability of a cloud to automatically expand or
compress the infrastructural resources on a sudden-up and down in the
requirement so that the workload can be managed efficiently. This
elasticity helps to minimize infrastructural cost. This is not applicable for
all kind of environment, it is helpful to address only those scenarios where
the resources requirements fluctuate up and down suddenly for a specific
time interval. It is not quite practical to use where persistent resource
infrastructure is required to handle the heavy workload.
It is most commonly used in pay-per-use, public cloud services. Where IT
managers are willing to pay only for the duration to which they consumed
the resources.

Example:
Consider an online shopping site whose transaction workload increases
during festive season like Christmas. So for this specific period of time,
the resources need a spike up. In order to handle this kind of situation, we
can go for Cloud-Elasticity service rather than Cloud Scalability. As soon
as the season goes out, the deployed resources can then be requested for
withdrawal.

Cloud Scalability:
Cloud scalability is used to handle the growing workload where good
performance is also needed to work efficiently with software or
applications. Scalability is commonly used where the persistent
deployment of resources is required to handle the workload statically.

Example:
Consider you are the owner of a company whose database size was small
in earlier days but as time passed your business does grow and the size of
your database also increases, so in this case you just need to request your
cloud service vendor to scale up your database capacity to handle a heavy
workload.
It is totally different from what you have read above in Cloud Elasticity.
Scalability is used to fulfill the static needs while elasticity is used to
fulfill the dynamic need of the organization. Scalability is a similar kind of
service provided by the cloud where the customers have to pay-per-use.
So, in conclusion, we can say that Scalability is useful where the workload
remains high and increases statically.

Types of Scalability:
1. Vertical Scalability (Scale-up):
In this type of scalability, we increase the power of existing resources in
the working environment in an upward direction.

188
188
 Cloud Issues and Challenges

2. Horizontal Scalability:
In this kind of scaling, the resources are added in a horizontal row.

3. Diagonal Scalability:
It is a mixture of both Horizontal and Vertical scalability where the
resources are added both vertically and horizontally.

12.2.3 Resource management and scheduling:

Cloud resource management requires complex policies and decisions for
multi-objective optimization. Effective resource management is extremely
challenging due to the scale of the cloud infrastructure and to the
unpredictable interactions of the system with a large population of users.
The scale makes it impossible to have accurate global state information
and the large user population makes it nearly impossible to predict the
type and the intensity of the system workload.
An overview of policies and mechanisms for cloud resource management
is followed by a presentation of energy efficiency and cloud resource
utilization and the impact of application scaling on resource management.

189
Distributed System and The policies for CRM can be loosely grouped into five classes:
Cloud Computing
(1) admission control;
(2) capacity allocation;
(3) load balancing;
(4) energy optimization; and
(5) quality of service (QoS) guarantees.
The explicit goal of an admission control policy is to prevent the system
from accepting workload in violation of high-level system policies. A
system should not accept additional workload if this would prevent it from
completing work already in progress or contracted. Limiting the workload
requires some knowledge of the global state of the system.
Capacity allocation means to allocate resources for individual instances.
An instance is an activation of a service on behalf of a cloud user.
Locating resources subject to multiple global optimization constraints
requires a search in a very large search space. Capacity allocation is more
challenging when the state of individual servers changes rapidly.
Load balancing and energy optimization are correlated and affect the cost
of providing the services; they can be done locally, but global load
balancing and energy optimization policies encounter the same difficulties
as the capacity allocation. Quality of service (QoS) is probably the most
challenging aspect of resource management and, at the same time,
possibly the most critical for the future of cloud computing.
Resource management policies must be based on a disciplined approach,
rather than ad hoc methods.
To our knowledge, none of the optimal or near-optimal methods to address
the five classes of policies scale up, thus, there is a need to develop novel
strategies for resource management in a computer cloud. Typically, these
methods target a single aspect of resource management, e.g., admission
control, but ignore energy conservation; many require very complex
computations that cannot be done effectively in the time available to
respond.
Performance models required by some of the methods are very complex,
analytical solutions are intractable, and the monitoring systems used to
gather state information for these models can be too intrusive and unable
to provide accurate data. Many techniques are concentrated on system
performance in terms of throughput and time in system, but they rarely
include energy trade-offs or QoS guarantees. Some techniques are based
on unrealistic assumptions; for example, capacity allocation is viewed as
an optimization problem, but under the assumption that servers are
protected from overload.

190
190
Virtually all mechanisms for the implementation of the resource Cloud Issues and Challenges
management policies require the presence of a few systems which monitor
and control the entire cloud, while the large majority of systems run
applications and store data; some of these mechanisms require a two-level
control, one at the cloud level and one at the application level. The
strategies for resource management associated with IaaS, PaaS,
and SaaS will be different, but in all cases the providers are faced with
large fluctuating loads.
In some cases, when a spike can be predicted, the resources can be
provisioned in advance, e.g., for Web services subject to seasonal spikes.
For an unplanned spike, the situation is slightly more complicated. Auto-
scaling can be used for unplanned spike loads provided that: (a) there is a
pool of resources that can be released or allocated on demand and (b) there
is a monitoring system which allows a control loop to decide in real time
to reallocate resources.

Policies and mechanisms for resource management

A policy typically refers to the principal guiding decisions, whereas
mechanisms represent the means to implement policies. Separation of
policies from mechanisms is a guiding principle in computer science.
Cloud resource management policies can be loosely grouped into five
classes:
1. Admission control.
2. Capacity allocation.
3. Load balancing.
4. Energy optimization.
5. Quality-of-service (QoS) guarantees.
The explicit goal of an admission control policy is to prevent the system
from accepting workloads in violation of high-level system policies; for
example, a system may not accept an additional workload that would
prevent it from completing work already in progress or contracted.
Limiting the workload requires some knowledge of the global state of the
system. In a dynamic system such knowledge, when available, is at best
obsolete. Capacity allocation means to allocate resources for individual
instances; an instance is an activation of a service. Locating resources
subject to multiple global optimization constraints requires a search of a
very large search space when the state of individual systems changes
rapidly.
Load balancing and energy optimization can be done locally, but global
load-balancing and energy optimization policies encounter the same
difficulties as the one we have already discussed. Load balancing and
energy optimization are correlated and affect the cost of providing the
191
Distributed System and services. Indeed, it was predicted that by 2012 up to 40% of the budget for
Cloud Computing IT enterprise infrastructure would be spent on energy .
The common meaning of the term load balancing is that of evenly
distributing the load to a set of servers. For example, consider the case of
four identical servers, A,B,C, and D, whose relative loads
are 80%,60%,40%, and 20%, respectively, of their capacity. As a result of
perfect load balancing, all servers would end with the same load -50% of
each server’s capacity. In cloud computing a critical goal is minimizing
the cost of providing the service and, in particular, minimizing the energy
consumption. This leads to a different meaning of the term load balancing;
instead of having the load evenly distributed among all servers, we want to
concentrate it and use the smallest number of servers while switching the
others to standby mode, a state in which a server uses less energy. In our
example, the load from D will migrate to A and the load from C will
migrate to B; thus, A and B will be loaded at full capacity,
whereas C and D will be switched to standby mode. Quality of service is
that aspect of resource management that is probably the most difficult to
address and, at the same time, possibly the most critical to the future of
cloud computing.
As we shall see in this section, often resource management strategies
jointly target performance and power consumption. Dynamic voltage and
frequency scaling (DVFS)1 techniques such as Intel’s SpeedStep and
AMD’s PowerNow lower the voltage and the frequency to decrease power
consumption.2 Motivated initially by the need to save power for mobile
devices, these techniques have migrated to virtually all processors,
including the ones used for high-performance servers.
As a result of lower voltages and frequencies, the performance of
processors decreases, but at a substantially slower rate [213] than the
energy consumption. Table 6.1 shows the dependence of the normalized
performance and the normalized energy consumption of a typical modern
processor on clock rate. As we can see, at 1.8 GHz we save 18% of the
energy required for maximum performance, whereas the performance is
only 5% lower than the peak performance, achieved at 2.2 GHz. This
seems a reasonable energy-performance tradeoff!

12.3 QUALITY OF SERVICE (QOS) AND RESOURCE

ALLOCATION
In a general sense, Quality of Service (QoS) is defined as a set of quality
requirements (i.e., desirable properties) of an application, which are not
explicitly formulated in its functional interfaces (1.4.1 ). In that sense,
QoS includes fault tolerance, security, performance, and a set of "ilities"
such as availability, maintainability, etc. In a more restricted meaning
(considered in this chapter), QoS characterizes the ability of an application
to satisfy performance-related constraints. This specific meaning of QoS is
specially relevant in areas such as multimedia processing, real-time
control, or interactive services for end users.
192
192
Performance control is achieved through resource management, the main Cloud Issues and Challenges
theme of this chapter. We examine the main abstractions and patterns for
managing resources, including the use of feedback control methods.

Introducing Resource Management:

The function of a computing system is to provide services to its users.
Each service is specified by a contract between a service provider and a
service requester. This contract defines both the functional interface of the
service and some extra-functional aspects, collectively known as Quality
of Service (QoS), which include performance, availability, security, and
need to be accurately specified for each application or class of
applications. The part of the contract that defines QoS is called a Service
Level Agreement (SLA). The technical expression of an SLA usually
consists of a set of Service Level Objectives (SLO), each of which defines
a precise objective for one of the specific aspects covered by the SLA. For
instance, for an SLA on the performance of a web server, an SLO can
specify a maximum response time to be achieved for 95% of the requests
submitted by a certain class of users.

Motivation and Main Definitions:

In order to perform its function, a computing system uses various
resources such as processors, memory, communication channels, etc.
Managing these resources is an important function, and there are several
strong reasons for performing it accurately:
● Maintaining quality of service: Various indicators related to QoS in
user applications, specially performance factors, are directly
influenced by resource allocation decisions.
● Resource accounting: The users of a shared facility should be
charged according to their actual resource consumption. Therefore,
any consumed resource must be traced back to a user activity.
● Service differentiation and resource pricing: The resource
management system may allow users to pay for improved service.
The system should guarantee this differentiated form of service, and
the pricing scheme should adequately reflect the added value thus
acquired.
● Detecting and countering Denial of Service (DoS): A DoS attack
aims at preventing useful work from being done, by undue massive
acquisition of resources such as CPU time, memory, or network
bandwidth.
● Tracking and eliminating performance bugs: Without accurate
monitoring of resource usage, a runaway activity might invisibly
consume large amounts of resources, leading to performance
degradation in user applications; or a regular activity may reserve
unnecessary resources, thus hampering the progress of other activities.

193
Distributed System and In the traditional view of a computing system, resource allocation, i.e., the
Cloud Computing sharing of a common set of resources between applications contending for
their use, was a task performed by the operating system, and user
applications had little control over this process. This situation has changed
due to the following causes:
● The increasing number of applications subject to strong constraints in
time and space, e.g., embedded systems and applications managing
multimedia data.
● The growing variability of the environment and operating conditions
of many applications, e.g., those involving mobile communications.
● The trend towards more open systems, and the advent of open
middleware.
Thus, an increasing part of resource management is being delegated to the
upper levels, i.e., to the middleware layers and to the applications
themselves. In this chapter, we examine some aspects of resource
allocation in these upper levels.
The term resource applies to any identifiable entity (physical or virtual)
that is used by a system for service provision. The entity that actually
implements service provision, using resources, is called a resource
principal. Examples of physical resources are processors, memory, disk
storage, routers, network links, sensors. Examples of virtual resources are
virtual memory, network bandwidth, files and other data (note that virtual
resources are abstractions built on top of physical resources). Examples of
resource principals are processes in an operating system, groups of
processes dedicated to a common task (possibly across several machines),
various forms of "agents" (computational entities that may move or spread
across the nodes of a network). One may define a hierarchy of principals:
for example, a virtual machine provides (virtual) resources to the
applications it supports, while requesting (physical or even virtual)
resources from a hypervisor.
Goals and Policies:
The role of resource management is to allocate resources to the service
providers (principals), subject to the requirements and constraints of both
service providers and resource providers. The objective of a service
provider is to respect its SLA, the contract that binds it to service
requesters (clients). The objective of a resource provider is to maximize
the utilization rate of its resources, and possibly the revenue it draws from
their provision. The relationships between clients, service providers, and
resource providers are illustrated in below figure,

194
194
 Cloud Issues and Challenges

Figure: Service providers and resource providers

The requirements are the following, as seen by a service provider.
● The service provider should offer QoS guarantees to its clients, as
specified by an SLA. The SLA may take various forms: strict
guarantee, probabilistic (the agreed levels will be reached with a
specified probability, or for a specified fraction of the demands), best
effort (no guarantee on the results). The SLA is a contract that goes
both ways, i.e., the guarantees are only granted if the clients respects
some conditions on their requests. These conditions may apply to
each client individually or to a population of clients as a whole.
● The service provider allocates resources for the satisfaction of clients'
requests. It should ensure equitable treatment (fair share), in the sense
that each client needing a resource should be guaranteed a share of
that resource proportional to its "right" (or "priority"), as defined by a
global policy. The share should be guaranteed in average over a
specified period of time. If all clients have the same right, then each
should be guaranteed an equal share. Other situations may occur:
● Service differentiation is a means of specifying different classes
of clients with different rights, which may be acquired by
purchase or by negotiation.
● The rights may be time-dependent, e.g., the right of a client may
be increased to allow it to meet a deadline.More generally, the
guarantee may be in terms of a minimal rate of progress, which
prevents starvation, a situation in which a client’s requests are
indefinitely delayed.
More generally, the guarantee may be in terms of a minimal rate of
progress, which prevents starvation, a situation in which a client's requests
are indefinitely delayed.
● If several classes of clients are defined, the service provider should
guarantee service isolation: the allocation of resources for a class
should not be influenced by that of other classes. This means that a
195
Distributed System and misbehaving client (one that does not respect its side of the SLA) may
Cloud Computing only affect other clients of its class, not clients from other classes.
The above requirements may be contradictory. For instance, ensuring
guarantees through worst case reservation may entail sub-optimal resource
utilization. Such conflicts may only be resolved according to a higher level
policy, e.g., through priority setting or through negotiation. Another
approach is to pool resources, in order to amortize their cost among
several applications, provided that peak loads do not occur at the same
time for all of them. Also note that the fairness requirement should be met
both at the global level (sharing resources among service providers) and
for each service provider with respect to its own service requesters.
Resource management policies may be classified using various criteria
(based on prediction and/or observation, using open loop or closed loop)
and may face various operating conditions: if resources are globally
sufficient to meet the global demand, combinatorial optimization is the
relevant tool; if resources are insufficient (the common case), control
methods are more appropriate. Examples throughout this chapter illustrate
these situations.
Resource allocation is subject to a number of known risks, which any
resource management policy should consider.
● Violation of fairness, examples of which are the above-mentioned
starvation, and priority inversion, a reversal of prescribed priorities
due to unwanted interference between synchronization and priority
setting.
● Congestion, a situation in which the available resources are
insufficient to meet the demand. This may be due to an inadequate
policy, in which resources are over-committed, or to a peak in the
load. As a result, the system's time is essentially spent in overhead,
and no useful work can be done, a situation known as thrashing.
Thrashing is usually avoided or delayed by an admission control
policy
● Deadlock, a situation of circular wait, in which a set of processes are
blocked, each of them waiting for a resource that is held up by another
member of the set. Deadlock may be prevented by avoiding circular
dependencies (e.g., through ordered allocation), or detected and
resolved, usually at some cost in progress rate.
In the context of this book, we are specifically interested in resource
management for middleware systems; among these, Internet services are
the subject of an intense activity, due to their economic importance. We
conclude this section with a review of the main aspects of resource
management for this class of systems.

196
196
12.1.3 Resource Management for Internet Services: Cloud Issues and Challenges

An increasing number of services are available over the Internet, and are
subject to high demand. Internet services include electronic commerce, e-
mail, news diffusion, stock trading, and many other applications. As these
services provide a growing number of functions to their users, their scale
and complexity have also increased. Many services may accept requests
from millions of clients. Processing a request submitted by a client
typically involves several steps, such as analyzing the request, looking up
one or several databases to find relevant information, doing some
processing on the results of the queries, dynamically generating a web
page to answer the request, and sending this page to the client. This cycle
may be shortened, e.g., if a result is available in a cache. To accommodate
this interaction pattern, a common form of organization of Internet
services is a multi-tier architecture, in which each tier is in charge of a
specific phase of request processing.
To answer the demand in computational power and storage space imposed
by large scale applications, clusters of commodity, low cost machines
have proved an economic alternative to mainframes and high-performance
multiprocessors. In addition to flexibility, clusters allow high availability
by replicating critical components. Thus each tier of a cluster-based
application is deployed on a set of nodes (Figure). How the application
components running on the servers of the different tiers are connected
together depends on the architecture of the application; examples may be
found in the rest of this chapter. Nodes may be reallocated between
different tiers, according to the resource allocation policy.

Figure: A cluster-based multi-tier application

A service provider may deploy its own cluster to support its applications.
An alternative solution, in increasing use, is for the provider to host the
application on a general-purpose platform (or data center) owned by a
computing facility provider, and shared with other service providers. The
drawback of this solution is that the service provider has less control on
the fine-grain tuning of the infrastructure on which its application is
running. However, there are several benefits to using shared platforms.
● The service provider is freed from the material tasks of maintaining
the infrastructure.

197
Distributed System and ● The fixed cost of ownership (i.e., the part of the cost that is not
Cloud Computing proportional to the amount of resources) is shared between the users
of the common facility.
● Mutualizing a large pool of resources between several applications
allows reacting to load peaks by reallocating resources, provided the
peaks are not correlated for the different applications.
● Resource sharing improves global availability, because of redundancy
in hosts and network connections.

12.4 IDENTITY AND ACCESS MANAGEMENT

Identity and Access Management (IAM) is a combination of policies and
technologies that allows organizations to identify users and provide the
right form of access as and when required. There has been a burst in the
market with new applications, and the requirement for an organization to
use these applications has increased drastically. The services and resources
you want to access can be specified in IAM. IAM doesn’t provide any
replica or backup. IAM can be used for many purposes such as you want
to secure the control of individual and group access from your AWS
resources. With IAM policies, managing permissions to your workforce
and systems to ensure least-privilege permissions becomes easier. The
AWS IAM is a global service.

Components of IAM
Users
Roles
Groups
Policies
With these new applications being created over the cloud, mobile and on-
premise can hold sensitive and regulated information, It’s no longer
acceptable and feasible to just create an Identity server and provide access
based on the requests. In current times an organization should be able to
track the flow of information and provide least privileged access as and
when required, obviously with a large workforce and new applications
being added every day it becomes quite difficult to do the same. So
organizations specifically concentrate on managing identity and its access
with the help of few IAM tools. It’s quite obvious that it is very difficult
for a single tool to manage everything but there are multiple IAM tools in
the market that help the organizations with any of the few services given
below.
Services By IAM:
Identity management
Access management
198
198
Federation Cloud Issues and Challenges
RBAC/EM
Multi-Factor authentication
Access governance
Customer IAM
API Security
IDaaS – Identity as a service
Granular permissions
Privileged Identity management – PIM (PAM or PIM is the same)

Figure : Services under IAM

More About the Services: Looking into the services on brief, Identity
management is purely responsible for managing the identity lifecycle,
access management is responsible for the access to the resources, access
governance is responsible for access request grant and audits, PIM or
PAM is responsible for managing all the privileged access to the
resources. The remaining services either help these services or help in
increasing the productivity of these services.
Market for IAM: Current situation of the market, there are three market
leader (Okta, Saipoint and Cyberark) who master one of the three domains
(Identity Management, Identity Governance and Privilege access
management), according to Gartner and Forrester reports. These
companies have developed solutions and are still developing new
solutions that allow an organisation to manage identity and its access
securely without any hindrances in the workflow. There are other IAM
tools, Beyond Trust, Ping, One login, Centrify, Azure Active Directory,
Oracle Identity Cloud Services and many more.
*****
199