Digital Assignment-1

Name: A.Rahitya Teja

Reg.No: 20BIT0296

Course Code: ITE3007

Course Name: Cloud Computing and Virtualization

NFS file services

Abstract:

NFS is a common file-sharing protocol that is used to efficiently access remote data. It
provides a versatile technique of remote file access by allowing a user to access remote data
as if it were local data. NFS excels in writing files with small record sizes, whereas iSCSI
excels at writing files with higher record sizes. Yet, there is no discernible difference between
NFS and iSCSI in reading activity. iSCSI is the better protocol to employ in an OpenStack
Cinder implementation for cloud storage. NFS and iSCSI can be configured in the Cinder
configuration file on the client side, and on the server side on a Synology NAS storage
system.
Introduction:

Data storage is the most important in the network system, file system is such a way that it has
the files which are named and logically arranged for doing the operations like accessing the
data[1]. The NFS(Network File System) was designed by Sun Microsystems, it is a client-
server application that it provides the shared file system across a network to the
clients[2].NFS provides transparent and remote access to the file systems. File servers are
transforming into very important in the modern network acting as a centralized data sharing
and storage stations[1]. Cloud computing is a term used nowadays. Cloud computing usage is
already spread all over the world and used by resources more effectively but also makes it
much easier to utilize a virtual machine. Several alternative protocols can many companies.
Because of its high usage, a need for cloud storage now emerges. Cloud storage not only uses
be implemented for cloud storage, and each protocol has its advantages. This research aims to
determine the most resources more effectively but also makes it much easier to utilize a
virtual machine[4].Cloud storage services, such as Dropbox, Google Drive, iCloud Drive, and
Microsoft OneDrive, have revolutionized the way people store and synchronize their files
across different devices. Dropbox stands out due to its delta sync feature, which provides
enhanced network-level security. These services are highly popular and greatly simplify the
task of accessing files from multiple devices[3].

The design and the implementation:

It consists of the three parts which are 1.protocol, 2. Server, 3. Client[1].

The NFS protocol uses a stateless design, which means that the server does not keep track of
the state of the client. This simplifies the design and implementation of the protocol but can
result in some performance limitations, such as the need to re-send data in the event of a
network error.

[1]
This is the basic architecture of the nfs which includes client and server

On the server side, In order to uniquely identify files on the server, each file is assigned a file
handle (handle) which includes the inode number, inode generation number, and filesystem
id. The fhandle is used by the client to reference the file in subsequent requests. The inode
number and generation number are used to identify the file, while the filesystem id is used to
distinguish the file on the server from other files that may have the same inode number and
generation number but reside on a different filesystem.
However, it is possible for a file to be deleted and then a new file created with the same inode
number and generation number, which can lead to confusion on the server if it receives a
request with an fhandle that references the original file. To handle this, the server must
maintain a cache of recently deleted inodes and check if the inode referred to by the fhandle
has been recently deleted and reused. If it has, the server can return an error to the client
indicating that the file no longer exists. This ensures that the server can correctly handle
fhandles even when the inode number and generation number are reused.

The Sun version of client-side NFS provides a transparent interface for remote file access on
the client side. To accomplish this, a technique of locating remote files that did not affect the
structure of path names was required, as some Unix-based solutions need changes to current
programmes. Instead of late binding, the hostname lookup and file address binding are
performed only once per filesystem, allowing the client to mount a remote filesystem to a
directory. This method has the advantage of requiring the client to deal with hostnames only
once, at mount time, and allowing the server to limit filesystem access by confirming client
credentials. The downside is that remote files are not accessible until a mount has been
completed. To provide transparent access to different types of filesystems mounted on a
single machine, a new filesystem interface was developed in the kernel that supports both a
Virtual Filesystem (VFS) interface and a Virtual Node (node) interface[7].

This is the overall view of the design and implementation.

Now let us look at the variations of the NFS

1.Replicated NFS and transparency
2.Parallel NFS
3. Mobile Agent Based NFS
4. Web Service Based NFS
5. gVault: Gmail based cryptographic NFS

Let us discuss one by one in detail:

1.Replicated NFS and Transparency:

The Ficus file system is unique in its ability to allow updates during network partition if any
copy of the file is accessible. Any updates made to files or directories are automatically
propagated to accessible replicas. Conflicting updates to directories are detected and repaired
automatically, while conflicting updates to ordinary files are reported to the owner. The
optimistic scheme of the Ficus file system is attractive due to the rarity of conflicting updates
and the infrequency of communication outages that render replicas inaccessible.
Additionally, the Ficus file system uses stackable layers to add new features to an existing
file system without having to re-implement existing functions. This is done by structuring the
file system as a stack of modules with the same interface, allowing for transparent
augmentation of existing services. The implementation of the Ficus file system using the
layered architecture is described in this paper [9].

TOFF aims to achieve two primary goals, namely replication transparency and failure
transparency. Replication transparency refers to the ability to replicate data without the
clients being aware of it or needing to maintain a list of replicated servers. In other words, the
replication process should be seamless and invisible to the clients. Meanwhile, failure
transparency pertains to the ability of the system to mask any failure, such that the clients are
not aware of the failure and can continue to access the system as if nothing happened. These
two goals are essential in ensuring a robust and reliable distributed system. [10].

(From google images)

2. Parallel NFS:

NFS protocol enables file sharing across multiple systems. pNFS extension was introduced to
improve performance by allowing the distribution of file access across multiple servers.
pNFS decouples the data path of NFS from its control path to allow concurrent I/0 streams
from multiple client processes to storage servers. This approach is unique when compared to
other parallel file systems like GPS, Panasas, Lustre, and PVFS. pNFS is the only open
standard that is being developed and approved by various commercial and nonprofit
organizations [11]. Parallel NFS (pNFS) is a protocol designed for parallel I/O access in
various storage environments. It has been validated through several prototypes and efforts to
improve its performance have focused on exposing the best bandwidth potential from
underlying file and storage systems. The pNFS architecture is discussed, and a proposed
direct I/O request flow protocol is suggested to further enhance its performance. Overall,
NFS is an emergent standard protocol with potential for efficient parallel I/O access.

3.Mobile Agent Based NFS:

MADFS is a new distributed file system designed to improve performance in WAN

environments by reducing network transfer overhead and cache management. It uses a
hierarchical host structure and mobile agents for transmission, communication, and
synchronization. HCCM is a novel cache coherency mechanism for MADFS that reduces
management overhead. Compared to conventional distributed file systems, MADFS with
HCCM shows better performance in WAN [12].
MADFS is a platform for mobile agent where agents handle transfer, communication, and
management tasks. There are three types of agents in MADFS: IA, WA, and DMA. IA runs
on client hosts and processes file system calls, WA executes file operations, and DMA
manages domain name, cache, property, space, and access control. MMA manages and
coordinates all DMAs in MADFS. Agents can move to target servers for better performance.
(Google)

Web Services based NFS:

The Web-services-based network file system (WSNFS) is a solution developed to overcome

the issues of file sharing between different file servers. A Web services file gateway is
installed to support this, allowing the client to mount all file directories on the file servers that
will be shared. The client makes file-operation requests to the gateway, which forwards each
request to the proper file server and returns the result to the client. The communication
between the client and the gateway adheres to the XML Web services standard, guaranteeing
that it is secure and dependable. This method addresses the difficulties associated with
sharing files across several file servers and offers the client with a unified experience. The
usage of a Web services file gateway also enables effective and optimised connection
between the client and the file servers.
Overall, the WSNFS provides an excellent solution for file sharing in a heterogeneous
environment, allowing enterprises to smoothly transfer data across many file servers. [13].
gVault: Gmail based cryptographic NFS:

The introduces gVault, a cryptographic network file system that uses Gmail's storage to
provide users with a free and easily accessible network drive. gVault offers benefits such as
secure remote access, 24/7 availability, and large storage capacity. The design and
implementation of gVault address challenges to ensure data confidentiality and integrity, with
a novel encrypted storage model and key management techniques. An initial prototype is
implemented and experiments show that the cost of security is negligible compared to data
transfer costs [15]. gVault operates as a client-side application that prompts users to enter
their Gmail login credentials and a master password. Once this information is entered, gVault
establishes a session with the Gmail server and operates as an HTTP client. All file
operations performed by the user are translated into equivalent HTTP requests and the
resulting HTML responses from the Gmail servers are parsed to retrieve the user's data. To
ensure the security of user data, gVault employs various cryptographic techniques in this
process. The gVault design assumes that everything outside the client machine's security
perimeter is untrusted, with no restrictions on potential attacks.

(from
Google images)

Conclusion:

In conclusion, the NFS protocol offers an efficient and flexible way to access remote data that
can be indistinguishable from local data to the user. Various implementations of NFS exist
for different situations and underlying communication technologies, making it one of the
most versatile methods of remote file access available today. Cloud sync services are
experiencing an upsurge in pop- ularity, which makes efficient service provisioning on both
client and server sides more and more crucial [3]. In terms of writing small record size files,
NFS has an advantage, while iSCSI is better for larger files. For reading, there is no
noticeable difference between the two protocols. Overall, iSCSI has a slight throughput
advantage, making it the preferred protocol for OpenStack Cinder. The client-side
configuration can be done through the Cinder configuration file, and the server-side
configuration can be done on Synology NAS storage for both NFS and iSCSI.
References:

[1] https://www.ijera.com/papers/vol 1 issue 3/ZW013721729.pdf

[2] http://www.news.cs.nyu.edu/~jinyang/fa08/papers/nfs033.pdf
[3] https://ieeexplore.ieee.org/document/7979973
[4] https://www.sciencedirect.com/science/article/pii/S187705092031139X
[5] https://www.researchgate.net/publication/260290925_Elsevier
[6] https://ojs.unud.ac.id/index.php/jik/article/view/6483
[7] https://www.cs.ucf.edu/~eurip/papers/sandbergnfs.pdf
[8] https://ieeexplore.ieee.org/abstract/document/546393
[9] https://www.isi.edu/people/johnh/PAPERS/Guy90b.pdf
[10] https://ieeexplore.ieee.org/document/546393
[11] https://ieeexplore.ieee.org/document/5470716
[12] https://ieeexplore.ieee.org/document/5209029
[13] https://www.researchgate.net/profile/Gwan-Hwan-
Hwang/publication/220587411_WSNFS_A_web-services-
based_network_file_system/links/0c960534934f3110bd000000/WSNFS-A-web-services-
based-network-file-system.pdf
[14] http://nozdr.ru/data/media/biblio/kolxoz/Cs/CsLn/D/Data and Applications Security 21
conf.(LNCS4602, Springer, 2007)(ISBN 9783540735335)(299s).pdf - page=169