0% found this document useful (0 votes)
78 views

Cloud Computing Delivery Architecture Models

Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
78 views

Cloud Computing Delivery Architecture Models

Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/221190836

Analysis of Cloud Computing Delivery Architecture Models

Conference Paper · March 2011


DOI: 10.1109/WAINA.2011.74 · Source: DBLP

CITATIONS READS
57 2,025

2 authors, including:

Irena Bojanova
National Institute of Standards and Technology
48 PUBLICATIONS   566 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Information Technology View project

Encyclopedia of Cloud Co... on Google View project

All content following this page was uploaded by Irena Bojanova on 14 November 2018.

The user has requested enhancement of the downloaded file.


2011 Workshops of International Conference on Advanced Information Networking and Applications

Analysis of Cloud Computing Delivery Architecture Models

Irena Bojanova Augustine Samba


Graduate School of Management and Technology College of Technology
University of Maryland University College Kent State University
Adelphi, USA Kent, USA
[email protected] [email protected]

Abstract—Cloud computing is one of the emerging technologies


that will lead to the next generation of Internet. It provides Services Sharing Cloud Computing
optimized and efficient computing through enhanced
collaboration, agility, scalability, and availability. In this Resource Sharing Grid Computing
paper, the evolution of the Cloud is discussed, the cloud
computing model is explained, a set of cloud computing Information Sharing The World Wide Web
infrastructure metrics is listed, the cloud service model
architectures are described, exemplary implementations of Network Sharing Inter-Networks
cloud service models are analyzed, and intriguing facts about The Internet
cloud computing status and future are shared. Networking
Networks
Keywords-Cloud Computing; SaaS; PaaS; IaaS; Google
Apps; Oracle Fusion; Microsoft Azure; Amazon EC2 Figure 1. The Evolution of Sharing on the Cloud.

According to the U.S. Government’s National Institute of


I. INTRODUCTION
Standards and Technology (NIST), cloud computing is a
The cloud computing paradigm enhances collaboration, “model for enabling convenient, on-demand network access
agility, scalability, and availability for end-users and to a shared pool of configurable computing resources (e.g.,
enterprises. It provides optimized and efficient computing networks, servers, storage, applications, and services) that
platform, and reduces hardware and software investment can be rapidly provisioned and released with minimal
cost, as well as carbon footprint . In this paper, the evolution management effort or service provider interaction” [2]. The
of sharing on the Internet is discussed, the cloud computing origin of the term comes from the early days of the Internet
model and the cloud delivery architecture models are where the network was depicted as a cloud (Fig. 2).
explained, exemplary implementations of cloud services are Cloud computing is an Internet-based delivery model for
analyzed, and intriguing reports and facts about the current Information Technology (IT) services that enhances
status of cloud computing and its future are shared. collaboration, agility, scalability, and availability. The
optimized and efficient computing is provided through a
II. EVOLUTION OF THE CLOUD virtualized technology infrastructure, which is maintained
The evolution of sharing on the Cloud (Fig. 1) went and secured for the users. Cloud computing services are
through: networking, network sharing, information sharing, offered on a pay-as-you-go basis and assure considerable
resources sharing, and services sharing [1]. The first stage of reduction in hardware and software investment costs, as well
the Cloud was around networking, the TCP/IP abstraction. as in carbon footprints and energy costs. Currently, cloud
Multiple regional networks, linking computers, were built at computing is more attractive for Web infrastructure,
universities and national laboratories. Their inter-networking
with TCP/IP led to network sharing and the emergence of
the Internet and its worldwide adoption. The second stage of
the Cloud was around documents, the WWW data
abstraction. The HTML format, the HTTP protocol, and the
Mosaic browser were adopted by universities for document Networks Services
exchange and then worldwide for information sharing. Then,
Servers THE CLOUD Storage
grid computing emerged with the creation of standards and
software for remote resources sharing and collaboration, Applications
exclusively utilized for highly scalable High Performance
Computing (HPC) jobs. The newest stage of the Cloud,
cloud computing, has emerged to provide services sharing
by abstracting infrastructure complexities of servers,
applications, data, and heterogeneous platforms.
Figure 2. The Standard Enterprise Cloud

978-0-7695-4338-3/11 $26.00 © 2011 IEEE 453


DOI 10.1109/WAINA.2011.74
collaboration, development and testing, high performance Hybrid clouds comprise unique clouds, bound by technology
computing applications, but still less attractive for database that enables interoperability amongst the clouds.
and transaction processing, and regulated applications. The relative efficiencies of the different cloud computing
models can be determined by measuring and analyzing the
III. THE CLOUD MODEL following set of cloud computing infrastructure metrics.
The Cloud model is composed of five essential x Hardware costs – measured by determining the
characteristics, three service models, and four deployment virtualized environment needed to consolidate a
models [2] as illustrated in Fig. 3. traditional infrastructure of independent servers. The
The essential cloud characteristics are on-demand self- consolidation goal is to increase the efficiency by
service, broad network access, resource pooling, rapid reducing the overall infrastructure costs through
elasticity, and measured service. Computing capabilities, elimination of redundant and under-utilized servers.
such as server time and network storage, can be unilaterally x Software costs – the dominant costs are determined
provisioned or de-provisioned as needed and automatically. by the suite of virtualization software and service
They are available over the Internet and accessible through management software. The virtualization software
heterogeneous client platforms, such as laptops and mobile provides the intelligence for server consolidation and
phones. The computing resources are pooled and improving system utilization by executing different
dynamically assigned and reassigned to serve multiple workloads on a single server. The service
consumers. The capabilities appear to be unlimited, as they management software facilitates efficient workflows,
can be rapidly and elastically provisioned to quickly scale administration and management of multiple
out and rapidly released to quickly scale in. The resource use workloads, and automated service provisioning.
is automatically controlled and optimized by leveraged x Real-time provisioning costs – measured by
metering capabilities. determining the elapsed time for deploying new
The cloud service models are Software as a Service systems, upgrading existing systems, or migrating to
(SaaS), Platform as a Service (PaaS), and Infrastructure as a a new consolidated cloud computing platform. Real-
Service (IaaS). SaaS allows use of applications that run on time provisioning is usually accomplished via
the cloud infrastructure and are accessible from various automated tools that may also be integrated into the
client devices. PaaS allows creation of applications with service management system.
supported programming languages and tools, and their x System administration costs – realized by measuring
deployment onto the cloud infrastructure. IaaS allows to the efficiency of managing virtual servers. Cloud
provision processing, storage, networks, and other computing models have more virtual servers and
fundamental computing resources to deploy and run comparably few physical servers. In addition, in
software, including operating systems and applications. today’s environment, virtual servers are more
With SaaS, the users do not manage or control even the complex to administer than physical ones. Effective
individual application capabilities. With PaaS they have system administration in the cloud computing model
control over the deployed applications and the hosting would require a paradigm shift from the traditional
environment configurations. With IaaS, users do not manage centralized approach to a fully distributed approach
or control the underlying cloud infrastructure, but they have neatly integrated into virtual servers.
control over the operating systems, deployed applications, In the current environment, clients of Cloud providers
storage, and limited control on networking components. can accrue significant reductions in overall IT costs by
The cloud deployment models are private, community, utilizing private cloud infrastructure compared to public
public, and hybrid. Private clouds are operated for individual cloud infrastructure. On the other side, Cloud providers
organizations. Community clouds are operated for accrue the most savings with public cloud infrastructure.
organizations with shared interests. Public clouds are A critical aspect of cloud computing is assessing the
available to the general public or large industry groups. risks, such as data integrity, recovery, privacy, and tenant
Essential Characteristics Cloud Service Models isolation, as the traditional requirements for IT security
On-demand Self-Service become more complicated by the externalization of the
Software as a Service (SaaS)
Broad Network Access Platform as a Service (PaaS) security model.
Resource Pooling Infrastructure as a Service (IaaS)
Rapid Elasticity IV. CLOUD SERVICE MODEL ARCHITECTURES
Measured Service
The cloud computing delivery architecture models define
Cloud Deployment Models the boundaries and the fit of the services, and how the set of
service models align and interact [3]. Fig. 4 illustrates the
Private Public entire stack of SaaS over PaaS over IaaS service models over
Cloud Cloud
Community Hybrid the cloud infrastructure, and the possible variations of SaaS,
Cloud Cloud PaaS, and IaaS architectures. Each *aaS model is a self-
contained solution stack of integrated functionality, where
Figure 3. The Cloud Model
IaaS provides the foundation and the other two – PaaS and
SaaS – in turn build upon IaaS cloud service model

454
Figure 4. The *aaS Stack and the Variations of Cloud Service Model Architectures.

architectures may be comprised by any logical combination facilitate communication, collaboration, business processes,
of the three *aaS service models over the cloud customer relationship management (CRM), enterprise
infrastructure. For example, a SaaS service model could resource planning, and human resources. Examples of the
stack directly over the cloud infrastructure, or a PaaS service SaaS paradigm include Google Apps, Microsoft Exchange,
model could stack over IaaS over the cloud infrastructure. Cisco WebEx Weboffice, Oracle CRM On Demand,
IaaS includes the entire infrastructure resource stack SalesForse.com, and Yahoo Mail.
(from facilities to hardware), has capability to abstract The Google SaaS cloud includes communication and
resources and deliver physical and logical connectivity to collaboration applications for end-users and organizations.
them, and provides application interfaces for interaction with Each tool is hosted by Google and offered on demand as
the infrastructure. PaaS is build upon IaaS, adding a layer of service instances to multiple users. Google Apps [4] tools
integration with application development frameworks include Gmail; Google Calendar for agenda management,
(middleware capabilities and functions), allowing developers scheduling, shared online calendars and mobile calendar
to build applications coupled to the platform. SaaS in turn is synchronization; Google Docs for documents, spreadsheets,
built upon the underlying IaaS and PaaS stacks, providing a drawings, and presentations sharing; Google Groups for
self-contained operating environment for delivery of the secure coding of free Web pages for intranets; and Google
entire user experience (content and its presentation, video for private, secure, hosted video sharing. Other Google
applications, and management capabilities). Apps include integrated applications built by independent
software developers and hosted at Google. The SaaS
V. EXPLORATION OF IMPLEMENTED CLOUD SERVICES presentation modality of Google Apps is a two steps process:
In the cloud computing paradigm, virtualization and the user initiates a download and then an install. The current
management of standard enterprise IT solution stacks are SaaS presentation platform supports: Microsoft Internet
offered through the three levels of services: SaaS, PaaS, and Explorer 7.0+, Mozilla Firefox 3.0+, Google Chrome 4.0+,
IaaS, which provide a composite view of cloud service and Safari 3.0+. The SaaS Application Programming
model architectures. Interface (API) is integrated into the Google universal
toolbar. The API therefore remains open while the browser is
A. Software as a Service running, and controls how the software application in the
In a SaaS cloud, a given application running at a data Cloud is harnessed by the end-user.
center is offered as service instances in real-time to several
end-users or organizations on demand. The basic SaaS B. Platform as a Service
architecture is included as part of the service model PaaS is typically a suite of low-level software, which
architecture illustrated in Fig. 4. SaaS applications may provides a platform for application-level development and

455
deployment. The basic PaaS architecture is included as part oriented architecture (SOA) technologies, and business
of the service model architecture illustrated in Fig. 4. It is process management (BPM). Above the middleware
offered as a service to developers to facilitate the complete typically there is a suite of APIs, which provide developers
software life cycle without the need to purchase standard with the capability to interact with the virtual platform
enterprise management tools and infrastructures (i.e., components (e.g. UI, SOA, and BPM) over the Cloud.
platform virtualization). The low-level software may include
application software, middleware, databases, and C. Infrastructure as a Service (IaaS)
development tools. Developers interface with the platform IaaS provides end-users and organizations with a suite of
through an API and a specific language (e.g. Java, C#, or virtual hardware and associated software as services over the
Python). The interface (e.g. API) provides developers with IaaS cloud. The basic IaaS architecture is included as part of
the capability to harness specific platform features provided the service model architecture illustrated in Fig. 4. Typical
as part of the virtualization over the Cloud. The platform virtual hardware includes servers, storage systems, routers,
features may include middleware and development tools, and switches. The associated software includes tools to
operating systems, and relational database services, present the virtual OS environment and file system
depending on the PaaS offerings. capabilities over the Cloud. End-users interface with the
Examples of PaaS offerings include Oracle Fusion services via a suite of APIs, provided as part of the IaaS
Middleware, Google AppEngine, Amazon Web Services, offerings. Examples of IaaS offerings include IBM, Amazon
Facebook, and Microsoft Azure. Fig. 5 illustrates the Elastic Compute Cloud (EC2), Microsoft Azure, Rackspace
fundamental architecture of a PaaS cloud for UI, SOA, and Cloud, Telstra, and Sun.
BPM service offerings. Note that Microsoft Azure capabilities are now extended
In particular, the Oracle Fusion Middleware PaaS to include both PaaS and IaaS clouds. Its framework starts
offering [5] includes user interface (UI) technologies, service with a set of high-availability machines, designated as the
“Fabric Controllers” (FCs) for the PaaS and IaaS clouds.
API: UI, SOA, BPM Provisioning Each FC is configured with both a Host and Guest Virtual
Machines (VM). Each VM includes Microsoft.Net
Cloud:
Framework, WindowsServer Internet Information Services
Virtual Management/ (IIS), and WindowsServer 2008 Enterprise. Fig. 6 illustrates
Middleware: UI SOA BPM Integration the basic architecture of Microsoft Azure PaaS and IaaS
service offerings.
Figure 5. Basic Architecture of PaaS

Figure 6. Basic Architecture of Microsoft Azure Platform.

456
Amazon Elastic Compute Cloud (EC2) is another includes Elastic and Block Store in designated availability
example of an IaaS public cloud model. Key components of zones for persistent high availability; and the Simple Storage
the architecture are Elastic Block Storage (EBS), Simple Service (SimpledB).
Storage Service (S3), Availability Zones and Regions, CPU
Cores, Virtual Machines, Virtualization software, Amazon VI. CONCLUSIONS
Machine Images, dynamic internal IP address, dynamic IT enterprises face challenges such as increasing costs and
external IP address schemes and fixed elastic IP addresses. constant need for capital investments; server sprawl, rising
Fig. 7 illustrates the basic architecture of Amazon EC2. operational costs, ballooning energy costs and demands,
In the Microsoft Azure environment, the Virtual increased complexity of systems and need for specialized
Machines are provisioned using declarative statements. talent to support them, variable recourse usage, demand for
Declarative descriptions of application components (i.e., business process simplification, time to market pressures.
Worker Roles) are also employed to implement various The embrace of cloud computing or IT as utility service
networking models. The storage platform includes SQL Data leads to more efficient use of IT hardware and software
Services and Azure Storage Service. investments, on demand provisioning or de-provisioning,
In the Amazon EC2 environment, the VM compute elastic scalability, usage based charging, service base
model is based on Xen Hypervisor VM. The VM can also be provisioning, and agility. The cloud computing idea is
provided via a third party, such as RightScale. The appealing to end-users and organizations because of its
networking model is implemented via declarative simplicity. There is no need to invest in expensive hardware
specifications of IP-address topology, including external and and software or pay for capacity that may not be used all the
elastic IP addresses; the availability zones are designed to time. The users have access anytime, through any device,
support high availability schemes. The storage model

Figure 7. Basic Architecture of Amazon Elastic Compute Cloud.

457
and from anywhere to software, platforms, and storage, applications development environments, and even
infrastructures, with no need to maintain and control them. infrastructure and security available from the Cloud.
Both established and start-up enterprises have shifted to
cloud computing; however there are still doubts about its REFERENCES
cost efficiency and robustness. McKinsey’s & Co. reported [1] J. Rymarczyk, “Virtualization for Data Centers of Today &
that cloud computing is beneficial for small and medium Tomorrow,” http://www-
businesses, but questioned the model sustainability for large 07.ibm.com/systems/includes/content/migratetoibm/pdf/
Virtualization_for_Data_Centers_of_Today_Tomorrow.pdf
corporations [6]. Outages and security concerns flawed the
[2] P. Mell and T. Grance, “The NIST Definition of Cloud Computing,”
prospects of cloud computing, too: power knock-out due to a v. 15, http://csrc.nist.gov/groups/SNS/cloud-computing
lightning strike crippled the Amazon EC2 cloud [3] CSA, “Security Guidance for Critical Areas of Focus in Cloud
infrastructure for hours [7], outages at Rackspace caused Computing,” V2.1,
numerous customer applications to go down [8], and a http://www.cloudsecurityalliance.org/guidance/csaguide.pdf
hacker distributed 300 confidential Twitter's business affairs [4] GoogleApps for Business,
documents [9], stored on Google Apps. In addition, http://www.google.com/apps/intl/en/business/index.html
according to Greenpeace, cloud computing may hold [5] Oracle Fusion Middleware,
sustainability implications as the cloud data centers are http://download.oracle.com/docs/cd/E15523_01/index.htm
“massive storage facilities that consume incredible amounts [6] McKinsey’s & Co, “Clearing the Air on Cloud Computing,”
of energy” [10]. http://images.cxotoday.com/cxoimages/storyimages/matter101157.pdf
Nevertheless, the concept of Cloud Computing is here to [7] Econsultancy, “Lightning strikes Amazon's EC2,”
http://econsultancy.com/blog/4013-lightning-strikes-amazons-ec2
stay and the cloud providers are making sure their cloud
[8] Rackspace Blog, http://www.rackspace.com/blog/?m=200912
services are safe. The governments realize the “flexibility,
operational benefits and substantial cost savings that cloud [9] Tweeter, “Tweeter CIO tweet on the hacking incident,”
http://twitter.com/ev/status/2647935038
computing can provide” [11]. For example, in May 2009, the
[10] Greenpeace, “Make IT Green: Cloud Computing and its Contribution
Japanese Government announced the Kasumigaseki Cloud to Climate Change,” The Netherlands: Greenpeace International,
[12]; in September 2009, the US Government launched the March `2010,
Cloud Computing Mall for government agencies [13], and in http://www.greenpeace.org/raw/content/international/press/reports/ma
January 2010, the UK Government introduced the G-Cloud ke-it-green-cloud-computing.pdf
government cloud infrastructure [14]. [11] CloudBook, “Government Clouds,”
Forrester Research reported that Infrastructure and http://www.cloudbook.net/directories/government-programs
Operations (I&O) “needs to focus the second half of 2010's [12] MIC, “Digital Japan Creation Project (ICT Hatoyama Plan),”
http://www.soumu.go.jp/main_sosiki/joho_tsusin/eng/Releases/News
budget on growth” by transforming the desktops, retooling Letter/Vol20/Vol20_01/Vol20_01.html
the data centers to support a path to the Cloud, and round out
[13] US Government Cloud Computing Mall, https://apps.gov
the efforts by industrializing the IT operations [15]. Cloud
[14] UK Government G-Cloud,
computing is no longer on the horizon; it has become the http://www.cabinetoffice.gov.uk/cio/ict/ict_strands/g_cloud.aspx
next logical step in enterprise computing. Organizations are [15] Forrester Research .Focus Your IO Budget On Three Key Initiatives;
focusing on managing information and no longer on http://access.forrester.com/rb/Research/focus_i%26o_budget_on_thre
managing infrastructure, by having their applications and e_key_initiatives/q/id/57354/t/2

458

View publication stats

You might also like