Cloud Computing Architecture

Ravi Shankar Dhakar1, Amit Gupta2, Ashish Vijay 3

1
.M Tech Student, Sri Balaji College of Engineering & Technology, Jaipur, Rajasthan
2
.Asstt. Profesor, Sri Balaji College of Engineering & Technology, Jaipur, Rajasthan
3
. M Tech Student, Sri Balaji College of Engineering & Technology, Jaipur, Rajasthan
ravidhakar83@gmail.com1, cseprof_amit@rediffmail.com2, ashishvijay2007@gmail.com3

Abstract- Cloud computing is Internet-based computing,
whereby shared resources, software and information are
provided to computers and other devices on demand. Users
dont need expertise in or control over the technology
infrastructure "in the cloud". Typical cloud computing
providers deliver common business applications online that are
accessed from another Web service or software like a Web
browser, while the software and data are stored on servers. The
two most significant components of cloud computing
architecture are front end and back end. The front end is the
part seen by the client, i.e. the computer user. This includes the
clients or computer and the applications used to access the
cloud via a user interface such as a web browser. The backend
of the cloud computing architecture is the cloud itself,
comprising various computers, servers and data storage
devices. A key element of cloud computing is customization and
creation of a user-defined experience.
Keywords: Cloud Infrastructure, deployment models, cloud
architecture.
I. INTRODUCTION
disillusionment among enterprise users. The Internet is
often represented as a cloud and the term cloud computing
arises from that analogy. McKinsey says that clouds are
hardware based services offering compute, network and
storage capacity where: hardware management is highly
abstracted from the buyer; buyers incur infrastructure costs
as variable OPEX [operating expenditures]; and
infrastructure capacity is highly elastic (up or down) [2].
Large companies can afford to build and expand their
own data centres but small- to medium-sized enterprises
often choose to house their IT infrastructure in someone
elses facility. A colocation centre is a type of data centre
where multiple customers locate network, server and storage
assets, and interconnect to a variety of telecommunications
and other network service providers with a minimum of cost
and complexity.
Amazon has a head start but well known companies
such as Microsoft, Google, and Apple have joined the fray.
[3]

Everyone has an opinion on what is cloud computing. It can
be the ability to rent a server or a thousand servers and run a
geophysical modelling application on the most powerful
systems available anywhere. It can be the ability to rent a
virtual server, load software on it, turn it on and off at will,
or clone it ten times to meet a sudden workload demand. It
can be storing and securing immense amounts of data that is
accessible only by authorized applications and users. It can
be supported by a cloud provider that sets up a platform that
includes the OS, Apache, a MySQL database, Perl,
Python, and PHP with the ability to scale automatically in
response to changing workloads. Cloud computing can be
the ability to use applications on the Internet that store and
protect data while providing a service anything including
email, sales force automation and tax preparation. It can be
using a storage cloud to hold application, business, and
personal data. And it can be the ability to use a handful of
Web services to integrate photos, maps, and GPS
information to create a mashup in customer Web browsers[1].
According to Gartners Hype Cycle Special Report for
2009, technologies at the Peak of Inflated Expectations
during 2009 include cloud computing, e-books and
Internet TV, while social software and micro blogging
siteshave tipped over the peak and will soon experience
Cloud Architectures are designs of software applications
that use Internet-accessible on-demand services.
Applications built on Cloud Architectures are such that the
underlying computing infrastructure is used only when it is
needed (for example to process a user request), draw the
necessary resources on-demand (like compute servers or
storage), perform a specific job, then relinquish the
unneeded resources and often dispose themselves after the
job is done. While in operation the application scales up or
down elastically based on resource needs.
Models-
Although not all the companies would agree on the
definitions given in this article, it is generally supposed that
there are three basic types of cloud computing: Infrastructure
as a Service (IaaS), Platform as a Service (PaaS) and
Software as a Service (SaaS).
II. RELATED WORK
In IaaS, cpu, grids or clusters, virtualized servers, memory,
networks, storage and systems software are delivered as a
service. Perhaps the best known example is Amazons
Elastic Compute Cloud (EC2) and Simple Storage Service
(S3), but traditional IT vendors such as IBM, and telecoms
providers such as AT&T and Verizon are also offering
solutions. Services are typically charged by usage and can
be scaled dynamically, i.e. capacity can be increased or
decreased more or less on demand.
PaaS provides virtualized servers on which users can
run applications, or develop new ones, without having to
worry about maintaining the operating systems, server
hardware, load balancing or computing capacity. Well
known examples include Microsofts Azure and Salesforces
Force.com. Microsoft Azure provides database and platform
services starting at $0.12 per hour for compute
infrastructure; $0.15 per gigabyte for storage; and $0.10 per
10,000 transactions. For SQL Azure, a cloud database,
Microsoft is charging $9.99 for a Web Edition, which
comprises up to a 1 gigabyte relational database; and $99.99
for a Business Edition, which holds up to a 10 gigabyte
relational database. For .NET Services, a set of Web-based
developer tools for building cloud-based applications,
Microsoft is charging $0.15 per 100,000 message
operations.
SaaS is software that is developed and hosted by the
SaaS vendor and which the end user accesses over the
Internet. Unlike traditional applications that users install on
their computers or servers, SaaS software is owned by the
vendor and runs on computers in the vendors data centre (or
a colocation facility). Broadly speaking, all customers of a
SaaS vendor use the same software: these are one-size-fits-
all solutions. Well known examples are Salesforce.com,
Googles Gmail and Apps, instant messaging from AOL,
Yahoo and Google, and Voice-over Internet Protocol (VoIP)
from Vonage and Skype.
2.1 CLOUD ARCHITECTURES
In traditional data processing it is difficult to get as many
machines as an application needs. Second, it is difficult to
get the machines when one needs them[5]. Third, it is
difficult to distribute and coordinate a large-scale job on
different machines, run processes on them, and provision
another machine to recover if one machine fails. Fourth, it is
difficult to autoscale up and down based on dynamic
workloads. Fifth, it is difficult to get rid of all those
machines when the job is done . Cloud Architectures solve
such difficulties. Applications built on Cloud Architectures
run in-the-cloud where the physical location of the
infrastructure is determined by the provider. They take
advantage of simple APIs of Internet-accessible services that
scale on demand, that are industrial-strength, where the
complex reliability and scalability logic of the underlying
services remains implemented and hidden inside-the-
cloud.[4] The usage of resources in Cloud Architectures is as
needed, sometimes ephemeral or seasonal, thereby providing
the highest utilization and optimum bang for the buck.
2.1.1 Business Benefits of Cloud Architectures
1. Almost zero upfront infrastructure investment: To build a
large-scale system, it may cost a fortune to invest in real
estate, hardware (racks, machines, routers, backup power
supplies), hardware management (power management,
cooling), and operations personnel. Now, with utility-style
computing, there is no fixed cost or startup cost.
2. Just-in-time Infrastructure: By deploying applications in-
the-cloud with dynamic capacity management, software
architects do not have to worry about pre-procuring capacity
for largescale systems. Cloud Architectures can relinquish
infrastructure as quickly as you got them in the first place (in
minutes).
3. More efficient resource utilization: System administrators
usually worry about hardware procuring and better
infrastructure utilization (when they have excess and idle
capacity). With Cloud Architectures they can manage
resources more effectively and efficiently by having the
applications request and relinquish resources only what they
need (on-demand).
4. Usage-based costing: Utility-style pricing allows billing
the customer only for the infrastructure that has been used.
The customer is not liable for the entire infrastructure that
may be in place. This is a subtle difference between desktop
applications and web applications.
5. Potential for shrinking the processing time: If one
compute-intensive or data intensive job that can be run in
parallel takes 500 hours to process on one machine, with
Cloud Architectures, it would be possible to spawn and
launch 500 instances and process the same job in 1 hour.
Having available an elastic infrastructure provides the
application with the ability to exploit parallelization in a
cost-effective manner reducing the total processing time. [6]
2.2 CISCO ARCHITECTURE
Cloud computing is a natural extension of the Cisco data
centre strategy [6]. Cisco has developed a roadmap of how
cloud data centres will evolve from the current state to an
eventual future state. In this multiphase roadmap, Cisco
walks through key cloud infrastructure evolution phases and
architectural enablers that Cisco brings to the enterprises and
the cloud computing industry.
Fig. 1Cisco Cloud Data Centre Evolution Road Map
2.2.1 Ciscos Solutions to Enable Cloud Data Centres
As part of the cloud data centre build-outs, Cisco is bringing
some key frameworks, architecture, and building blocks to
the industry.
2.2.2 Cisco Cloud Reference Architecture Framework
Cisco has developed the depiction of a cloud reference
architecture model, which portrays the architectural layers,
connected via APIs and repositories. If we study the
framework more closely, the following aspects can be
articulated. At the foundation of this framework is the data
centre technology architecture, which consists of three
salient blocks of network, compute, and storage. This layer
hosts all the services that are delivered to a cloud consumer
or subscriber. The next important layer is the security layer.
The key takeaway in this layer is that security is blanketed
as an end-to-end architecture across all aspects of the
framework. Security is considered as one of the key
challenges to be solved in a cloud framework; hence, it has
to be accounted for in a comprehensive sense. Following the
technology and security layer is the Service Orchestration
layer, which is implemented with configuration repository
enablers. The configuration repository stores key
information such as service catalogue, asset inventory, and
resource-to-service mappings. This layer is an important
layer because it maps the technology components to the
service components and serves as a reference point during
service provisioning. The service orchestration layer is the
glue that integrates the lower layers to create a service for
delivery. The next layer is also where infrastructure and
service management function take place. The topmost layer
is the consumer facing layer, usually exposed via a portal-
like solution. This is the layer where service is defined,
requested, and managed by the consumer. Cisco provides
not only this framework, but also key solutions to deliver
cloud services.
Fig.2 Cisco Cloud Reference Architecture Framework
2.2.3 Cisco Cloud Data Centre Building Blocks
Cisco brings an important set of technology building blocks
at the foundation of the cloud architectures. They are
described as follows.
2.2.3.1 10 Gigabit Ethernet
A cloud data centre is designed with the high density of
virtual machines coupled with a high processor core count.
From a networking perspective, multiple virtual machines
on a single server can quickly overwhelm a single Gigabit
Ethernet link, and multiple Gigabit Ethernet links can
increase costs. Moreover, there needs to be a strategy in
place to not only take the existing investment in 1 Gigabit
Ethernet and seamlessly integrate it into a 10 Gigabit
infrastructure, but to also enable migration to 10 Gigabit
Ethernet and unified fabric (described next). Cisco is
bringing new terminology and implementation to the
industry in this regards, called VN-Link. VN-Link is the
virtual link between the virtual machine and the physical
interface of the physical server. This implementation will
enable operational consistency down to the individual virtual
machine as well as policy portability, so network and
security policy follows virtual machines as they move
around the data centre. Specific benefits of Ciscos VN-Link
include:
Real-time policy-based configuration
Mobile security and network policy, moving policy with
the virtual machine during virtual machine mobility, and live
migration for persistent network, security and storage
compliance
Nondisruptive management model, aligning management
and operations environments for virtual machines and
physical server connectivity in the data centre
2.2.3.2 Unified Fabric
To continue to reduce the total cost of ownership (TCO) and
to deploy virtual machines, all servers must have a
consistent and ubiquitous set of network and storage
capabilities in a cloud data centre. One of the simplest and
most efficient ways to deliver these capabilities is to deploy
a unified fabric. The shift to a unified fabric gives all servers
(physical and virtual) access to the LAN, SAN, and IPC
networks, allowing more to be consolidated in the
customers network for greater efficiency and costs savings.
Cisco is offering not only 10 Gigabit Ethernet, but also
lossless 10 Gigabit Ethernet, currently called Data Centre
Ethernet or Enhanced Ethernet. This becomes the foundation
to consolidate fabrics like Fibre Channel (for SAN), which
require the stringent lossless nature of a network.
Additionally, a new breed of adapters, called converged
network adapters (CNAs), would be implemented in the
server platform, which will act at the consolidation and
virtualization point in the compute layer.
2.2.3.3 Unified Computing
The unified fabric now enables a fully virtualized
cloud data centre with pools of computing, network, and
storage resources, through the Cisco Unified Computing
System (UCS). The Cisco UCS bridges the silos in the
classic data centre, enabling better utilization of
infrastructure in a fully virtualized environment, and creates
a unified architecture using industry-standard technologies
that provide interoperability and investment protection. UCS
unites computing, network, storage access, and virtualization
resources into a scalable, modular design that is managed as
a single energy-efficient system. This system is managed
through an embedded management framework in the Cisco
UCS platform.
2.3 SUN AND CLOUD COMPUTING
For more than a decade, Sun has been advancing the
state of the art of large-scale computing infrastructure that
has formed the foundation of cloud computing. Beginning in
the 1990s, Sun has been a leader in helping service providers
implement their large-scale networks so that they can serve
up to millions of customers. Large numbers of Sun servers
are employed by financial institutions and stock exchanges
to handle tasks ranging from handling transactions to real-
time fraud detection. Sun has been a leader in high-
performance computing by developing rack-at-a-time
deployment models, automatic provisioning from the bare
metal to the application, and large scale virtual networking
with one petabit per second throughput. Indeed, Sun
pioneered the clouds predecessor, grid computing, by
selling physical server time by the hour and helping
customers implement internal grids to support their own
operations. Just as much as these large-scale computing
capabilities help Sun develop cloud computing solutions,
they are done in the context of advancing the systemic
qualities that these solutions require: scalability, availability,
reliability, manageability, and security.
2.3.1 Innovations from the Sun community
Sun has developed foundational technologies for cloud
computing and has been a central player in the community
development processes they have promoted. While Sun has
long maintained the Solaris Operating Systems industry
leadership, it has also spawned a corresponding open source
movement around the OpenSolaris Operating System. The
MySQL database is the Web application database of choice,
and the Java programming language powers Web sites and
enterprise data centres worldwide. The community-based,
open source GlassFish application server provides a Java
software execution container that has been extended to
support Ruby applications and the Drupal content
management system. OpenSolaris Project Crossbow has
helped to expand the multi-tenancy support in Sun xVM
hypervisor. Beneath the rich, community-supported
Software that Sun helps to foster comes the powerful server,
storage, and networking products that make them perform
including standard, scalable x86-architecture servers,
Suns UltraSPARC processor powered server product line,
and servers that incorporate Suns energy-efficient, chip
multithreaded (CMT) UltraSPARC T1, T2, and T2 Plus
processors. Suns CMT processors process high-throughput
workloads so efficiently that they are harnessed in content
load balancing and application delivery products such as the
Zeus Extensible Traffic Manager. Suns Open Storage
products combine open source software with industry
standard hardware to help reduce reliance on high-priced,
purpose-built systems. Indeed, Suns ground-breaking Sun
Fire X4500 Server helped the industry see the benefits of
combining server and storage technology in the same
system. Sun delivers virtual networking for large-scale
computing through InfiniBand to massive-scale compute
grids with the Sun Datacenter Switch 3456, scaling up to
13,834 nodes.
2.4 AMAZON CLOUD ARCHITECTURE: GREP THE WEB
The Alexa Web Search web service allows
developers to build customized search engines against the
massive data that Alexa crawls every night. One of the
features of their web service allows users to query the Alexa
search index and get Million Search Results (MSR) back as
output. Developers can run queries that return up to 10
million results. This application is currently in production at
Amazon.com and is code-named GrepTheWeb because it
can grep (a popular Unix command-line utility to search
patterns) the actual web documents. GrepTheWeb allows
developers to do some pretty specialized searches like
selecting documents that have a particular HTML tag or
META tag or finding documents with particular
punctuations,or searching for mathematical equations ,
source code, e-mail addresses or other patterns such as
(dis)integration of life. The output of the Million Search
Results Service, which is a sorted list of links and gzipped
(compressed using the Unix gzip utility) in a single file, is
given to GrepTheWeb as input. It takes a regular expression
as a second input. It then returns a filtered subset of
document links sorted and gzipped into a single file. Since
the overall process is asynchronous, developers can get the
status of their jobs by calling GetStatus() to see whether the
execution is completed. Hence, the design goals of
GrepTheWeb included to scale in all dimensions (more
powerful pattern-matching languages, more concurrent users
of common datasets, larger datasets, better result qualities)
while keeping the costs of processing down. The approach
was to build an application that not only scales with demand,
but also without a heavy upfront investment and without the
cost of maintaining idle machines (downbottom). To get a
response in a reasonable amount of time, it was important to
distribute the job into multiple tasks and to perform a
Distributed Grep operation that runs those tasks on multiple
nodes in parallel. It uses the following AWS components:
Amazon S3 for retrieving input datasets and for storing the
output dataset
Amazon SQS for durably buffering requests acting as a
glue between controllers
Amazon SimpleDB for storing intermediate status, log, and
for user data about tasks
Amazon EC2 for running a large distributed processing
Hadoop cluster on-demand
Hadoop for distributed processing, automatic
parallelization, and job scheduling.
Fig. 3 Grap The Web Architecture-Zoom Level 1
In the next four subsections we present rationales of
useand describe how GrepTheWeb uses AWS services.
In GrepTheWeb, Amazon S3 acts as an input as
well as an output data store. The input to GrepTheWeb is the
web itself (compressed form of Alexas Web Crawl), stored
on Amazon S3 as objects and updated frequently. Because
the web crawl dataset can be huge (usually in terabytes) and
always growing, there was a need for a distributed,
bottomless persistent storage. Amazon S3 proved to be a
perfect fit.
Amazon SQS was used as message-passing
mechanism between components. It acts as glue that wired
different functional components together. This not only
helped in making the different components loosely coupled,
but also helped in building an overall more failure resilient
system.
One use for a database in Cloud Architectures is to
track statuses[7]. Since the components of the system are
asynchronous, there is a need to obtain the status of the
system at any given point in time. Moreover, since all
components are autonomous and discrete there is a need for
a query-able datastore that captures the state of the system.
Because Amazon SimpleDB is schema-less, there is no need
to define the structure of a record beforehand. Every
controller can define its own structure and append data to a
Job item. For example: For a given job, run email
address regex over 10 million documents, the launch
controller will add/update the launch_status attribute
along with the launch_starttime, while the monitor
controller will add/update the monitor_status and
hadoop_status attributes with enumeration values
(running, completed, error, none). A GetStatus() call will
query Amazon SimpleDB and return the state of each
controller and also the overall status of the system.
Fig. 4 Grap the Web Architecture-Zoom Level 2
In GrepTheWeb, all the controller code runs on
Amazon EC2 Instances. The launch controller spawns
master and slave instances using a pre-configured Amazon
Machine Image (AMI). Since the dynamic provisioning and
decommissioning happens using simple web service calls,
GrepTheWeb knows how many master and slave instances
needs to be launched. Hadoop Map Reduce Hadoop is an
open source distributed processing framework that allows
computation of large datasets by splitting the dataset into
manageable chunks, spreading it across a fleet of machines
and managing the overall process by launching jobs,
processing the job no matter where the data is physically
located and, at the end, aggregating the job output into a
final result. In Hadoop, theres a master process running on
one node to oversee a pool of slave processes (also called
workers) running on separate nodes. Hadoop combines all
the results and sorts the results by the keys. The master then
assigns keys to the reducers. The reducer pulls the results
using an iterator, runs the reduce task (logic specified by
user), and sends the final output back to distributed file
system.
2.5 PROPOSED WORK
Ciscos next-generation cloud data architecture is based on a
unique, unified, and integrated approach, which addresses
these specific facets of cloud computing. With the
virtualized environment defined by a dynamic, scalable data
center fabric, a workload really can run anywhere; the
resources needed to support a workload can come even from
an outside service provider in a cloud-computing model.
Thus Cisco brings not only concepts and technologies, but
also key standards and partnerships to tackle these
challenges.
A beauty of GrepTheWeb is its almost-zero
infrastructure before and after the execution. The entire
infrastructure is instantiated in the cloud triggered by a job
request (grep) and then is returned back to the cloud, when
the job is done. Moreover, during execution, it scales
ondemand; i.e. the application scales elastically based on
number of messages and the size of the input dataset,
complexity of regular expression and so-forth.
 Sun innovations are the foundational technologies for
cloud computing environments that are open, standards-
based, and are the fruit of a community effort. Joining the
Sun cloud computing community means having the choice
of server, storage, and networking technologies that work at
maximum scale. Sun can help organizations build their own
private, local clouds as a way to transition enterprise
datacenters toward this new computing model while
retaining the utmost control over business-critical data. Sun
can help companies build their own private, non-local clouds
as a way to leverage the low cost of new large,
energyefficient colocation facilities such as Switch
Communications SuperNAP facility in Las Vegas, Nevada.
Sun can help those wishing to become cloud providers with
the hardware, software, and management capabilities that
are required.
II.CONCLUSION
Instead of building the applications on fixed and rigid
infrastructures, Cloud Architectures provide a new way to
build applications on on-demand infrastructures. Without
having any upfront investment, it is possible to run a job
massively distributed on multiple nodes in parallel and scale
incrementally based on the demand (users, size of the input
dataset). With no idle time, the application infrastructure is
never underutilized. As per the Cisco vision, cloud service
providers will offer both public and private cloud services in
the form of virtual private clouds that will be consumed and
controlled within private cloud data centers. Organizations
everywhere can augment their private clouds with Suns
public cloud offering either by collocating with Sun at the
SuperNAP site and enjoying the benefits of high-speed,
local infrastructure, or by using Suns services over the
Internet. Good cloud architectures should be impervious to
reboots and re-launches[8]. In GrepTheWeb, by using a
combination of Amazon SQS and Amazon SimpleDB, the
overall controller architecture is more resilient. If the
instance on which controller thread was running dies, it can
be brought up and resume the previous state as if nothing
had happened. This was accomplished by creating a pre-
configured Amazon Machine Image, which when launched
dequeues all the messages from the Amazon SQS queue and
their states from the Amazon SimpleDB domain item on
reboot.
REFERENCES
[1]. Jason Carolan and Steve Gaede, Introduction to Cloud Computing
architecture, SUN Microsystems Inc., pp. 1-40, June 2009
[2]. McKinsey & Co. Report presented at Uptime Institute Symposium,
Clearing the Air on Cloud Computing,April 18, 2009
[3]. Clash of the Clouds. The Economist October 15, 2009.
http://www.economist.com/displaystory.cfm?story_id=14637206
(accessed November 15, 2009).
[4]. Farber, R. Cloud Computing: Pie in the Sky? ScientificComputing.com,
November/December 2009
[5]. Jinesh Varia, Cloud Architectures, Amazon Web Services, pp.- 1-14
[6]. Kapil Bakshi, Cisco Cloud Computing-Data Centre Strategy,
Architecture and Solutions, U.S. Public Sector, Cisco, pp. -1- 16
[7]. Proffitt, A. Pharmas Early Cloud Adopters. BioIT World,
November/December 2009, pp.31-32.
[8]. Describing the Cloud. Details of CambridgeSofts Hosted Computing
Environment. accessed November 28, 2009.