Application Architecture for Cloud Computing

1
Unit-4
Application
Architecture for Cloud

2
Application Architecture
• What is Application Architecture?
• Application Architecture is the design of a software application
that outlines internal subcomponents and interaction between
modules and interface with external applications or services.
• It is a design or plan that defines what the application will
contain and how it will interact with infrastructure components.
• Why?
• The application is designed to automate specific business tasks in
a coherent and logical manner to help users conveniently
interact with others to store and share data.

3
Application Architecture?
• Why Cloud Architecture Required?
• In traditional application architecture, one or more Web
servers interact with the database using a middle tier
software or application framework.
• Traditional architecture is not scalable and not meant for
huge variations in user or system load.

4
Cloud Application Requirements
• Without a documented design and plan, cloud developers
will fail to capitalize on the advantages of cloud over
traditional environments and on cloud practices and
patterns.
• The new applications must be able to coexist with and use
other cloud services such as a cloud-based authentication,
security, and replication.
• While working with cloud applications, requirements and
architecture must be the first two documents to be written
and reviewed.

5
• There are two types of requirements: functional and non-
functional.
• Functional requirements list the purpose and objectives
of the application.
• Non- functional requirements include performance,
response time, built-in security, replication, ease-of- use,
productivity, agility, backups, business continuity,
scalability and modularity.
• These requirements are shown in Figure 1:

6
Application Requirements &
Constraints
• Business needs
• Required outcome Enterprise Vision
• Enterprise Vision
• Legal Limitations when hosting in
cloud
• Regulatory Requirements
• Cloud standards
• Use of Existing templates
• Corporate Policies for cloud use
Non-Functional Requirements
• Performance & Response Time
• Service availability in the cloud
• Security
• Backup to other clouds
• Extension to Hybrid clouds
• Localization
• Compatibility with other cloud
platforms Support for end-user devices
Functional
Requirements
• Required Features
• Business goals
• User requirements

7
Fundamental Requirements for Cloud
Application Architecture
• Here are a few practices for application architecture for clouds:
1. Cloud Applications Must be Flexible, Dynamic, and Distributable:-
2. Cloud Applications Must be Architected and Installed for Unknown
and Varying Geographic Locations:-
3. Cloud Applications Must Account for Pricing of Resource Access and
Utilization:-
4. Cloud Applications Must Take Care of Data Integrity and Consistency-
5. Cloud applications must process various information trypes:-
6. Cloud applications must be more Mobile-Aware:-
7. Applications must do lot more than just accepting and storing input:-

8
Relevance and Use of Client Server
Architecture for cloud Applications
• The developers have to write following two applications:
1. Server application for the cloud:-
2. Client Application for the Client Devices:-

9
Relevance and Use of Client Server
Architecture for cloud Applications
• Cloud vendors provide Integrated Development
Environment (IDE) so that programmers working on the
platform can create cross-platform browser-based as well
as rich out-of browser applications.

10
Addressing Cloud Application
Performance and Scalability
• A common implementationfor web based
application is the three tier architecture. It consist of:
1. A front end web server providing static content and some
cached dynamic content. It is the client access software or
code used to access the application.
2. A middle tier to provide dynamic content processing
via an application or development framework such as Ruby
on Rails, Java, .NET or open-source based application.
3. A back-end database management system that stores and
provides access to the user information.

11
• Data transfer and interaction between different tiers is part
of the architecture.
• Several sockets and protocols are used such as SNMP,
(simple network management protocol) Web services,
CORBA and UDP.(user datagram protocol)
• Cloud application servers can use scale-up or scale-out
mechanisms to add resources to meet workload demands.

12
• Scale-up Architecture:- Scale-up refers to instances when you
add more resources within an existing application server to meet
the needs.
• For example, you add memory or CPU within the same physical or
virtual machine.
• The advantage is that you do not need to add more overhead or
framework.
• You only need to identify the bottle-neck and add resources to
address the problem-at-hand.
• The problem is that a heavily-loaded server may be under
utilized most of the time when user loads are low.

13
• Scale-out Architecture:- Scale-out architecture adds more
processing power but in a different manner.
• The application is written so that it splits the user load over multiple
servers.
• Each server manages a small part of the overall work.
• This architecture offers the benefit of using multiple servers to
process the work.
• The environment has high fault tolerance.
• If a server fails, another node is available to take over, thus
alleviating availability concerns. IF more servers are required, one
can be taken from another pol, which has unused servers.

14
Service Oriented Architecture (SOA)
for Cloud Applications
• Implementation of cloud architecture requires a set of
principles and design patterns.
• It provides developers with guidance and direction.
• It helps them reduce the risks, costs and time associated with
building, deploying and managing an application that can
successfully deliver the desired business value.
• One of the best set of architecture guidelines for cloud
applications is defined within what is called Service Oriented
Architecture (SOA). It is a set of methodology to design a cloud
application in the form of interoperable units or services.

15
• These services can be reused for various other purposes
within the cloud.
• Other cloud developers are free to use and combine these
services to create new applications.
• Another key feature of SOA is that the functionalities or
services are unassociated or behave as loosely coupled
units.
• Each service is developed to implement a single action.

16
• A key benefit is that the SOA architecture can be used to
support communication between services.
• The communication can involve data transfer, exchange
information on state of application users, or coordinate to
perform an activity.

17
• Each interaction between services is independent of other
interactions.
• In that sense, they are self-contained. Since the ultimate
results are tied to a common user application, these
services are loosely coupled with each other.
• Another benefit of SOA is the modularity. Large application
development

18
• SOA enables large applications to be broken into smaller
components.
• They can be developed independently.
• Each of these smaller components is referred to as a service.
• Later these components are assembled or loosely coupled to
meet business needs.
• A SOA application can be said to a modular, loosely coupled set
of services designed to meet a business need.
• Since the services or components can be ported to another
platform, they have high cross-platform interoperability.

19
• Because a cloud is a set of services utilizing resources from a
virtual, dispersed pool, SOA components or services are ideal
for deployment in a cloud.
• Cloud developers usually associate individual SOA objects with
functionality in a non-hierarchical manner. They commonly
use a cloud-based software tool or catalog that contains all the
available services, their features and a process to assemble it
to help build a cloud application.
• Before cloud computing, SOA principles have widely been
used for distributed computing and modular
programming.

20
• Now it has become an operative architecture for cloud
based SaaS services.
• The various common interaction patterns used for SOA are
shown in Figure 2.
Resource - Oriented
SOA
● REST
(Representation
al State
Transfer)
● WOA (Web-
Method-oriented SOA
● SOAP
(Simple
Object Access
protocol)
● WS (Web
Event-Driven
SOA

21
• This is not to say that al SOA applications will look and act
the same. Figure 2 shows the three interaction patterns
that can be used in SOA deployments:
• Resource Oriented SOA leverages the architecture of the
Web and Web standards (e.g., HTTP and URIs) to scale
adoption and performance of the cloud applications.
• It uses Representational State Transfer (REST) Web
services.
• This architecture has been used to design large-scale public
clouds.

22
• Method-Oriented SOA uses Simple Object Access Protocol (SOAP)
based Web services standards.
• It helps provide common request/reply interactions (between
service provider and service consumer programs) to cloud
developers who use different development tools or middleware.
• The Web Services Description Language (WSDL) is commonly
used to describe the service, the SOAP protocol is used to
exchange structured data during the implementation of Web
services (WS) in the cloud.
• It uses Extensible Markup Language (XML) for its message format
and uses HTTP and SMTP for message transmission.

23
• Event driven SOA is based on the asynchronous exchange of message
amongst applications and user devices.
• The cloud application receives a message about an event as soon as it
is generated and published by the source program.
• The events or messages are analyzed and used in real-time
dashboards.
• These event-driven approaches have proved to be critical in creating
dynamic cloud applications and solutions that depend on pattern-
matching and context-based automation.
• This is of great value for real-time decision-making especially for sales
teams, customer contact centers, and supply chain management.

24
• There are certain remarkable benefits of event-driven
cloud applications to business and technical managers.
• It allows business executives to make tactical and
transactional decisions based on up-to–date information
and deep insight into relevant context for the decision.
• For making strategic decisions, they can use near-term,
comprehensive data from the application.
• Businesses can use the pattern-matching features while
processing large amounts of incoming data to find telling
patterns within buyer and seller preferences.

25
• These preferences can be used for real-time initiatives such
as making context-based pricing and sourcing decisions.
• For a technical manager, the loose coupling between
different components enables effective reuse.
• It also makes the applications highly flexible to take
advantage of the elasticity in the cloud.
• Each of these approaches must demonstrate the core
characteristics of SOA services:- i.e., they must be
modular, distributable, loosely coupled, swappable and
discoverable.

26
• At the same time, these contextual factors will influence
decision making around application architecture:
• In the near future, EDA will be briskly accepted, as the
ability to process contextual events and integrate
applications with data-collecting devices improves.
• This integration will be utilized to enhance cloud
application dashboards with more context-rich interfaces.
• These will provide the status of business and help make
business decisions with updated information.

27
• Because of the highly-valuable nature of pairing events, context and
actions, event driven models will be embedded in many new cloud
applications, and this architectural style will be adopted by many
organizations developing SOA based applications.
• The scope and scale of data exchange between users and organizations
are rapidly expanding.
• In many ways, the focus of SOA initiatives has shifted from internal
applications to external organizations (partners, customers).
• Web oriented Architecture (WOA) fits the inter-organization situations
more readily than traditional SOA based applications. Many SOA styles
will transition to use WOA, primarily due to the large number and
proportion of external facing services.

28
• As per Gartner, in their Article ID number G00175166:
• More than 60% of SOA projects had a positive impact on
their organizations’ ability to grow revenue.
• SOA projects generate positive returns in relatively short
periods, typically within 10 months.
• SOA improves the agility of the IT organization and the
overall enterprise.
• SOA can reduce the cost of building IT systems, but the cost
reduction are often indirect.

29
Parallelization within Cloud
Applications
• The on-demand availability of a large amount of processing
capability and memory in the cloud force architects to use
resources judiciously.
• Traditional architectures are based on availability of horizontal
scaling for front-end and middle tier and vertical scaling for the
back-end tier. However, these assumptions are not true for the
cloud.
• There is massive horizontal scaling on the back-end and middle-
tier.
• The logical separation between the tiers becomes intermingled
and hazy.

30
Applications
• The data processing must therefore be treated differently.
• The large number of user devices necessitates an expanded
middle-tier, which must, in parallel, analyze a lot of incoming
data. This requires instantaneous access to multiple back-end
systems.
• The need to manage multiple requests, requiring enormous
processing, forces the developer to parallelize the architecture
and send requests to multiple processors.
• The challenge is that the only real form of parallelism in many
applications is to have separate user requests operate in parallel
on different processors.

31
Applications
• The existence of massive memory helps to reduce context
switching, which in turn helps support multiple users.
• However, with real-time business analytics and user
context-sensitive data processing being done in the cloud,
single requests need to use several processors in parallel.
This architecture needs to be built into the application.

32
Leveraging In-Memory Operations for
Cloud Applications
• The use of memory is another architecture enhancement.
• Traditionally, application architects have treated database
systems as file stores for putting in and taking out table subsets.
• Data storage layers are used for integrity checking and data
validation.
• All data processing was done in the application memory.
• Over the years, database designs have been highly turned and
became very efficient.
• They have stored procedures that can push processing into the
database.

33
Cloud Applications
• Cloud applications working with large data chunks must
use in-memory processing on database servers.
• It creates the opportunity to use data management in more
innovative and productive ways.
• It can be used to create various in-memory layouts to
implement highly parallel processing of the data.
• Since the database has specific knowledge of location and
layout, it can optimize complex functions for the middle-
tier.

34
Cloud Applications
• The in-memory data management provides incredible speeds for
complex tasks such as business analytics, user location and context-
sensitive processing of data.
• The advantage of in-memory data management is, that “faster is
better”, but that the real-time results can be provided to the users for
active decision-making , rather than after-the-fact information to help
future decisions.
• With the cloud being used from mobile devices to access data, as and
when required, the users expect the applications to provide real-time
results at any instant.
• In-memory data management needs to be a central part of the cloud
application architecture.

Application Architecture for Cloud Computing

Recommended

More Related Content

Similar to Application Architecture for Cloud Computing (20)

Recently uploaded (20)

Application Architecture for Cloud Computing