cocsi-india.org.in/communications/csic may 2011.pdf · 2019. 1. 15. · csi report : csi discover...

Congratulations for the new format of CSI Communications!

The new editors on board are always tasked with the responsibility of change with continuity and you have

managed the transition very effi ciently. I have always been arguing that CSI

Communications should be differently positioned from CSI Journal with less academic/ research bias. Our President M.D. Agarwal is a crusader in the cause of making the Communications a journal for industry users as well. This has been ably achieved.

S Ramanathan, Senior CSI Life Member

The layout, color and over all it is excellent!

The CSI Communications layout, format, color, brief articles are very good. Overall the get up is excellent. Best wishes!

Bhagavan Sastry, CSI MemberIt looks great!

I just had a look at the printed CSI Communications April issue...and it looks

great! Congratulations to you and the entire team – at

the HQ and outside – who worked for this...!

Satish Babu, Vice President, CSI

Congratulation!!

I have received CSI Communications of April, 2011. It has given good feeling to

read this issue. The layout, content and fl ow of information have reached to a level

of international magazine. We hope to have more vibrant CSI now.

Azimuddin Khan , Vice Chairman, CSI Udaipur ChapterManager Systems, RSMM Ltd.

Amazing truely to international standard!

The April edition of CSIC is just amazing truely to international standard.

I congratulate whole editorial board. I can understand the amount of efforts that have undergone.

B. Tirumala Rao, Associate Professor, ANITS,Visakhapatnam

Congrats!

I can feel the difference in the April issue. It is surely markedly better.I like some of the following changes -• The articles are interesting,

much better and relevant• Call for Articles and Call for

proposals for events openly, is really nice and is more organized

• I like the divisions of the magazine into different sections - Prog. section, CIO etc.

I hope you are able to take the magazine to a different level.

Dr. Anita Goel, Reader, Dyal Singh College, University of Delhi

Excellent in format and content!

Congratulations to all! The new CSIC in print form is excellent in format and content. Keep up the good work.

V L Mehta, Director, MIEL

Hard work!

I would really like to appreciate the hard work and innovation put by the CSI Editorial team.

Anurag Jagetia, Assistant Professor (IT), Bhilwara

Coo

mI have

s

gCoooonnn

rerefl oww

aI con

ddd

or

-

Share your feedback and views with CSIC ReaderSpeak().Email: [email protected]

ReaderSpeak( )

CSIC May 2011.indd B 5/9/2011 2:36:06 PM

CSI Communications | May 2011 | 1

ContentsVolume No. 35 • Issue No. 2 • May 2011

Cover Story 6 Platforms for Building and Deploying Applications for Cloud Computing Rajkumar Buyya and Karthik Sukumar

Technical Trends12 Quality of Service Design in Clouds Praveen Ganghishetti and Rajeev Wankar

Research Front16 Resource Management on Clouds: Handling Uncertainties in Parameters and Policies Shikharesh Majumdar

Article20 Cloud Networking Masum Z. Hasan

Practitioner Workbench22 Programming.Tips() Object Oriented Ramp up from Procedural – An Example Debasish Jana

23 Programming.Learn(“Perl”) Achuthsankar S. Nair & Parul Tyagi CIO Perspective24 IT Strategy Nuances IT Strategy Management Anil V Vaidya

27 Managing Technology Panel Discussion with Gautam Shroff, Venguswamy Ramaswamy & V. Srinivasa Raghavan from Tata Consultancy Services Dr. R M Sonar and Mrs. Jayshree A Dhere

HR

30 Getting Ready for Job Seeking Viju Swaminathan

CSI Communications

Editorial Board

Chief EditorDr. R M Sonar

EditorsDr. Debasish JanaDr. Achuthsankar Nair

Resident EditorMrs. Jayshree Dhere

AdvisorsDr. T V GopalMr. H R Mohan

Published byExecutive Secretary Mr. Suchit GogwekarFor Computer Society of India

Please note:CSI Communications is published by Computer Society of India, a non-profi t organization. Views and opinions expressed in the CSI Communications are those of individual authors, contributors and advertisers and they may differ from policies and offi cial statements of CSI. These should not be construed as legal or professional advice. The CSI, the publisher, the editors and the contributors are not responsible for any decisions taken by readers on the basis of these views and opinions.

Although every care is being taken to ensure genuineness of the writings in this publication, CSI Communications does not attest to the originality of the respective authors’ content.

© 2011 CSI. All rights reserved.

Instructors are permitted to photocopy isolated articles for non-commercial classroom use without fee. For any other copying, reprint or republication, permission must be obtained in writing from the Society. Copying for other than personal use or internal reference, or of articles or columns not owned by the Society without explicit permission of the Society or the copyright owner is strictly prohibited.

P L U SICT@Society : Healthy Computing : Body Achuthsankar S. Nair

32

Brain TeaserDebasish Jana

33

Ask an ExpertDebasish Jana

34

Happenings@ICTH R Mohan

35

On the Shelf! – Book reviewJayshree Dhere

36

SIG - View Point : Cloud Computing: Opportunity or RiskRavi Eppaturi

37

CSI Report : CSI Discover Thinking Quiz 2011Ranga Rajagopal

39

CSI Report : 2011 Lean Six Sigma International ConventionUjwal Tripurari

40

ExecCom Transacts 41

CSI News 43

CSIC May 2011.indd 1 5/9/2011 2:36:24 PM

CSI Communications | May 2011 | 2 www.csi-india.org

Know Your CSI

Executive Committee (2011-12/13) »

President Vice-President Hon. SecretaryMr. M D Agrawal Mr. Satish Babu Mr. H R [email protected] [email protected] [email protected]

Hon. Treasurer Immd. Past PresidentMr. V L Mehta Mr. P [email protected] [email protected]

Nomination Committee Members (2011-2012)

Prof. (Dr.) A K Nayak Mr. P R Rangaswami Mr. Sanjay K Mohanty

Regional Vice-Presidents

Region - I Region - II Region - III Region - IVMr. R K Vyas Prof. Dipti Prasad Mukherjee Mr. Anil Srivastava Mr. Sanjay MohapatraDelhi, Punjab, Haryana, Himachal Assam, Bihar, West Bengal, Gujarat, Madhya Pradesh, Jharkhand, Chattisgarh, Pradesh, Jammu & Kashmir, North Eastern States Rajasthan and other areas Orissa and other areas inUttar Pradesh, Uttaranchal and and other areas in in Western India Central & Southother areas in Northern India. East & North East India Eastern India

Region - V Region - VI Region - VII Region - VIIIProf. D B V Sarma Mr. C G Sahasarabudhe Mr. Ramasamy S Mr. Jayant KrishnaKarnataka and Andhra Pradesh Maharashtra and Goa Tamil Nadu, Pondicherry, International Members Andaman and Nicobar, Kerala, Lakshadweep

Division Chairpersons

Division-I : Hardware (2011-13) Division-II : Software (2010-12) Division-III : Applications (2011-13)Dr. C R Chakravarthy Dr. T V Gopal Dr. S [email protected] [email protected] [email protected]

Division-IV : Communications Division-V : Education and Research(2010-12) (2011-13)Mr. H R Mohan Dr. Manohar [email protected] [email protected]

Structure & Organisation http://www.csi-india.org/web/csi/structureNational, Regional & http://www.csi-india.org/web/csi/structure/nsc State Students CoordinatorsStatutory Committees http://www.csi-india.org/web/csi/statutory-committees Collaborations http://www.csi-india.org/web/csi/collaborations Join Now - http://www.csi-india.org/web/csi/joinRenew Membership http://www.csi-india.org/web/csi/renewMember Eligibility http://www.csi-india.org/web/csi/eligibilityMember Benefi ts http://www.csi-india.org/web/csi/benifi tsSubscription Fees http://www.csi-india.org/web/csi/subscription-feesForms Download http://www.csi-india.org/web/csi/forms-downloadBABA Scheme http://www.csi-india.org/web/csi/baba-schemePublications http://www.csi-india.org/web/csi/publicationsCSI Communications* http://www.csi-india.org/web/csi/info-center/communicationsAdhyayan* http://www.csi-india.org/web/csi/adhyayanR & D Projects http://www.csi-india.org/web/csi/47Technical Papers http://www.csi-india.org/web/csi/technical-papersTutorials http://www.csi-india.org/web/csi/tutorialsCourse Curriculum http://www.csi-india.org/web/csi/course-curriculumeNewsletter* http://www.csi-india.org/web/csi/enewsletterCurrent Issue http://www.csi-india.org/web/csi/current-issueArchives http://www.csi-india.org/web/csi/archivesPolicy Guidelines http://www.csi-india.org/web/csi/helpdeskEvents http://www.csi-india.org/web/csi/events1President’s Desk http://www.csi-india.org/web/csi/infocenter/president-s-deskExecCom Transacts http://www.csi-india.org/web/csi/execcom-transacts1News & Announcements archive http://www.csi-india.org/web/csi/announcements

CSI Divisions and their respective web links Division-Hardware http://www.csi-india.org/web/csi/division1Division Software http://www.csi-india.org/web/csi/division2Division Application http://www.csi-india.org/web/csi/division3Division Communications http://www.csi-india.org/web/csi/division4Division Education and Research http://www.csi-india.org/web/csi/division5

List of SIGs and their respective web linksSIG-Artifi cial Intelligence http://www.csi-india.org/web/csi/csi-sig-aiSIG-eGovernance http://www.csi-india.org/web/csi/csi-sig-egovSIG-FOSS http://www.csi-india.org/web/csi/csi-sig-fossSIG-Software Engineering http://www.csi-india.org/web/csi/csi-sig-seSIG-DATA http://www.csi-india.org/web/csi/csi-sigdataSIG-Distributed Systems http://www.csi-india.org/web/csi/csi-sig-dsSIG-Humane Computing http://www.csi-india.org/web/csi/csi-sig-humaneSIG-Information Security http://www.csi-india.org/web/csi/csi-sig-isSIG-Web 2.0 and SNS http://www.csi-india.org/web/csi/sig-web-2.0SIG-BVIT http://www.csi-india.org/web/csi/sig-bvitSIG-WNs http://www.csi-india.org/web/csi/sig-fwnsSIG-Green IT http://www.csi-india.org/web/csi/sig-green-itSIG-HPC http://www.csi-india.org/web/csi/sig-hpcSIG-TSSR http://www.csi-india.org/web/csi/sig-tssr

Other Links -Forums http://www.csi-india.org/web/csi/discuss-share/forumsBlogs http://www.csi-india.org/web/csi/discuss-share/blogsCommunities* http://www.csi-india.org/web/csi/discuss-share/communitiesCSI Chapters http://www.csi-india.org/web/csi/chapters

Important links on CSI website »

* Access is for CSI members only.

Important Contact Details »For queries, announcements, correspondence regarding Membership, contact [email protected] any other queries, contact [email protected]

CSIC May 2011.indd 2 5/9/2011 2:36:34 PM


The winds of transformation have started blowing over IT Industry. Veteran Banker, Shri K.V. Kamath was named as Chairman of Infosys Technologies, the second largest software company in India. With this it is clear that leadership in IT companies is now no longer going to be restricted to IT professionals. Emergence of new services and solutions for inclusive sector has started showing an impact on the growth of IT industry. Government of India has earmarked crores of rupees for empowering its citizens and the society at large by arranging innovative ICT solutions in various projects like UID and organising various schemes for skill development, enhancing of knowledge and education, upgrading infrastructure etc. Recently concluded CSI National Conference for “ICT FOR SOCIAL EMPOWERMENT OF RURAL INDIA” held at Tata Institute of Social Sciences, Mumbai“ did enlighten about new upcoming opportunities for IT industry. Demand of ICT solutions for inclusive growth has surfaced in education, agriculture, health etc., and it offers excellent opportunities in the areas of R&D and provides for the development of capability and competency for new emerging sectors.

Latest trends and developments in IT industry provide a good basis for planning various CSI programs and seminars and forming policies for our society. Question arises as to how these trends will have an impact on CSI plans; in what manner we should participate in the nation building exercises; what should be our strategy and action plan be to support these developments; and what kind of technical programs and seminars should be organized for better relevance and connectivity with Nation’s priorities.

Corporate user segment is looking at Cloud Computing to achieve economies of scale and agility in delivering their services to Customers. This emerging model now in its nascent stage in India, is posing a challenge to the conventional way of managing IT Services with in-house IT.

I am glad to mention that we are discussing the subject of cloud computing and related developments in this issue of CSI communications. This is an excellent opportunity for CSI to connect with the user community by arranging a good number of programs in this domain.

CSI recently launched CSI Communications with varied contents and new design, which was highly appreciated by our members. We are indeed proud of our editorial team consisting of Prof. Rajendra Sonar, Dr. Debasish Jana, Prof. Achuthsankar Nair and Ms. Jayshree Dhere, and our publishing team viz. Suchit, Shashank Kadam and our printer, who have professionally designed and brought forth a fully transformed CSI Communications. I am confi dent that with the high quality of technical content, CSIC will very soon match the category of world class publications of ACM, IEEE and will be of great help to improve the perception and brand image of CSI.

Our Education Directorate with the assistance of editorial team consisting of CSI veteran, Mr. HR Mohan, Dr. Vipin Tyagi and Dr. R.K. Vyas, has rolled out Student Newsletter “CSI Whizkidd“. I am in no doubt that soon this newsletter will evolve and do a good service to student members.

Another important aspect of our publication is CSI “Computer Science Research Journal”. Research Publication viz. Transactions is being rolled out with determined efforts of Prof. S V Raghavan and Mr. S. Mahalingam. Since, Project “CSI Transactions” may take another year, strategically, we have initiated a process for starting an online, General Purpose Research publication “CSI Research Journal”. My sincere gratitude to our Life member, research veteran & TIFR senior professor, Prof. Shyamasundar, who agreed to lead this activity as Chief Editor. Soon an editorial team will come into existence and plans will be known thereafter. Prof. Dipti Prasad Mukherjee, RVP, has laid the foundation of its revival.

I consider publications and periodicals an important service to members. High quality, good variety of content and consistency of these publications are great contributors to ratings and visibility/perception of society. Since last couple of years, we observed that there is not much growth of publications. Hence, for few months, till this activity gets stabilized, I have requested ExecCom that I would like to personally supervise as ex-offi cio Chairman of Publication Committee and try to build teams and long term plans. We need to outline our policies and schemes for publications for a horizon of 2-3 years. For this, it is important that we have a good team, partnership model with publishers and clarity of market demands, good understanding & linkages to research community, collaborations with premier research Labs like IBM, Intel, Infosys and engagement with senior academicians from IITs, IIITs & others and also build up a good operation and organization committee to support this. These Publications can fetch decent revenue for society.

Friends, I am concerned with the state of affairs at most of our Chapters. While student activities and technical programs are increasing manifold, Chapter level activities are fading, except few places like Mumbai, Hyderabad, Bangalore, Kolkata, Coimbatore and Chennai. We are not able to attract professionals. Our membership among IT professionals and corporate circles also needs much improvement.

President’s Message M. D. AgrawalFrom : [email protected] : President’s DeskDate : 1st May, 2011

Dear Friends

Trends and Opportunities

Publications and Research

Organization Issues

CSIC May 2011.indd 3 5/9/2011 2:36:34 PM


In my understanding, profi le & commitment of team members & their linkage with stake holders, clarity of ideas, planning and organization create a difference. We need to seriously create a balance team structure of representative stakeholders, CIOs, self-employed CEOs, Practitioners, Academicians, Govt. offi cials, IT professionals in each Committee, through nomination, if not through election route, under provision of co-option.

There is need for continuous mentoring and preparing agenda of activities. I would request RVPs, the Membership Committee & Director Education to spare time to co-ordinate this area. We shall have few location specifi c common programs. We shall not lose sight of our engagement with IT professionals and corporate in deciding these programs. Balance shall be maintained between programs useful for educational institutions, corporate and government. Good learning experience can be taken from Mumbai Chapter, which is successfully running useful programs.

CSI has introduced new scheme of bulk membership for corporate, the promotion of which requires persuasion in each major chapter.

With efforts of Dr. Bhatia & of some of RVPs, quite a few chapters are being revamped. My good wishes to them.

There is a need to have an organization strategy to identify program for addressing on-going issues.

It is time for Research and Development Sector to make a Paradigm shift to cope up with transformations that are taking place at Industry and Academic Levels. Former President Dr A.P.J Abdul Kalam has expressed his view that modern methods of research based learning and teaching can play an important role in development of the society - “The world in the 21st century will be a knowledge based society with multiple opportunities. What worked yesterday, won’t work today. Educational Institutions need to be the power houses for knowledge.”Promotion of research as career may be a good thrust area for student members. I am happy to note that our all-India student co-ordinator, Mr. Ranga Raj Gopal has started dialogue with team of regional RSC/ SSC. Our RVPs may start using similar approach with chapters for helping them in their activities. I recommend introduction of some meaningful programs for skill and competency development through certifi cation programs and try to run “Teach R & D“ kind of motivation workshops at some institutions both for faculty and students, with an objective of promoting research as an exciting career. A small team for this purpose may be formed by Director Education, Prof. N.L. Sarda, IIT, Mumbai, who is ready to help.Our Research Committee convener, Prof. Rajiv Sangal is trying to bring under CSI banner, few research conferences, which are currently being organized by individual groups in various parts of the country. We shall welcome such initiatives of collaboration and alignments with resourceful groups.New strategy for revival of SEARCC, which requires consistent efforts, was discussed during visit of SEARCC President, Mr. Anthony and President of Sri Lanka and PNG. CSI may play an important role in SEARCC.We need to vitalize SIGs functionality. SIGs are created for carrying out technical programs and promotion of research in each domain. We still have to realize our objectives. Some of SIGs are not able to do any program. There is need to look for a new approach and demand commitments from each SIG convener. We may consider formation of new committees for these non-functional SIGs, in next 3 to 6 months time. I convey my sincere appreciation to organizers of following events that took place during the month of April, 2011.

� NCVESCOM – 11: 4th National Conference on VLSI, Embedded Systems, Signal Processing & Communication Technologies, Chennai.

� National Conference on ICT for social Empowerment of Rural India, Mumbai. � International Conference on Emerging Trends in Networks & Computer Communications, Udaipur. � National Conference on Rural Trends in Computational Intelligence, Indore. � CDCT 2011: National Seminar on Convergence of Database and Communication Technologies. � Regional Student Convention, Hyderabad, held in March, 2011 – Region-V.

To conclude, I look forward to active participation by my colleagues in ExecCom, chapters, SIGs & Special committees to fully get involved in carrying out CSI Mission of Nation Building Initiatives.

Entire CSI family was deeply grieved by the loss of our beloved CSI Fellow, Dr. Chandwani. He will be known as the most caring CSI Veteran, foremost Educationist, Research Scholar and a highly dedicated and sincere person. I pay my sincere homage to Dr. Chandwani.

Jai hind!

With love,

M D Agrawal

“Complexity of a system is directly proportional to the simplicity of its original design.” – Anonymous.

Student Activities

CSIC May 2011.indd 4 5/9/2011 2:36:37 PM


Editorial Rajendra M Sonar, Achuthsankar S Nair, Debasish Jana and Jayshree DhereEditors

Dear Fellow CSI Members:

It gives us great pleasure in bringing out this issue to you with special focus on Cloud Computing, which has become the latest buzzword in computing.Cloud computing represents a new paradigm with an architectural shift from traditional distributed computing. Cloud computing presents computing resources like hardware, software, platform or infrastructure as utility oriented services to the consumers. Similar to electricity, gas, and water, computing is offered as services. The service offerings on virtualized platforms are aimed to be accessible anywhere with complete management by the provider, as on-demand, subscription based as well as elastic in terms of scalability. Improved resource utilization, higher independence on device and location, reduction in cost are some of the claimed benefi ts. Several open issues and concerns at the service and architecture levels include security, privacy, trust, integration standards, interoperability, availability, sustainability, dynamic cost model and loss of control over sensitive data. While cloud

computing is an active area of research, some of the prominent cloud service providers include Amazon (Web Services, Elastic Compute Cloud, Simple Storage Services), Microsoft (Azure), Salesforce.com, Google (Apps), Manjrasoft (Aneka).Vivek Kundra, the fi rst CIO of the USA, appointed by President Obama, summarized cloud computing service offereings as: “There was a time when every household, town, farm or village had its own water well. Today, shared public utilities give us access to clean water by simply turning on the tap; cloud computing works in a similar fashion. Just like water from the tap in your kitchen, cloud computing services can be turned on or off quickly as needed. Like at the water company, there is a team of dedicated professionals making sure the service provided is safe, secure and available on a 24/7 basis. When the tap isn’t on, not only are you saving water, but you aren’t paying for resources you don’t currently need.”We present the cover story written by Prof. Rajkumar Buyya of The University of Melbourne, Australia and Mr. Karthik Sukumar of Manjrasoft Pvt. Ltd, Australia. Through their lucid presentation, they have demonstrated about the Manjrasoft Aneka as a Cloud Application Platform utilizing the underlying concepts and allowing an easy development of cloud ready applications on a private, public or hybrid cloud infrastructure. The Research Front section presents a research direction towards effi cient cloud resource management under uncertainties written by Prof. Shikharesh Majumdar of Carleton University, Canada. Mr. Praveen Ganghishetti and Dr. Rajeev Wankar of University of Hyderabad present a Quality of Cloud Service Management Strategy in Technology Trends section. An interesting article is presented in Article section by Dr. Masum Z Hasan of Cisco Systems, San Jose,

USA. He focuses on few aspects of cloud networking including on-demand network management functions and interfaces, VM-aware networking and layer L3 scaling. A crossword puzzle tests your knowledge on cloud computing as the new paradigm.CIO Perspective provides you with a panel discussion on paradigm of cloud computing, its adoption, care to be taken while adopting and so on, with three eminent technologists from TCS, India’s biggest software company. Dr. Gautam Shroff, VP and Head, TCS Innovation Labs – Delhi, Venguswamy Ramaswamy, Head, the iON initiative and V. Srinivasa Raghavan, Cloud Computing initiative owner within the TCS CTO R&D Organization answer various questions that are posed to them on the specifi c topic of Cloud Computing from the viewpoint of CIOs, CTOs and CXOs handling IT.CIO Perspective also provides you with an article on IT Strategy, second in the series on IT Strategy Nuances, by Dr Anil Vaidya. The article focuses on IT Strategy Management and Execution and stresses its importance from the point of view of realizing the benefi ts and measuring the success factor. HR column provides useful tips to those who are getting ready for seeking a job. Tips are provided on different aspects such as resume writing, dressing up oneself for an interview and then presenting at the time of interview.Special Interest Group of CSI (CSI-SIG) on Cloud Computing provides a write-up on their views on Opportunities and Risks in connection with Cloud Computing. Mr. Ravi Eppaturi, Vice Chairman of CSI Mumbai Chapter has written this article. In addition to this there are regular features such as Practitioner Workbench, ICT@Society, Ask an Expert, On the Shelf (Book Review), Happenings@ICT and CSI News.

We are very thankful to all those, who have sent their feedback on the new format of the CSI Communications. Some of these have been compiled in a separate section called ReaderSpeak(). Do keep providing comments and feedback for further improvements.

With warm regards,

Rajendra M Sonar, Achuthsankar S Nair, Debasish Jana and Jayshree DhereEditors

“There was a time when every household, town, farm or village had its own water well. Today, shared public utilities give us access to clean water by simply turning on the tap; cloud computing works in a similar fashion. Just like water from the tap in your kitchen, cloud computing services can be turned on or off quickly as needed. Like at the water company, there is a team of dedicated professionals making sure the service provided is safe, secure and available on a 24/7 basis.”

Cloud computing represents a new paradigm with an architectural shift from traditional distributed computing.

CSIC May 2011.indd 5 5/9/2011 2:36:38 PM


Cover Story

Platforms for Building and Deploying Applications for Cloud Computing

Rajkumar Buyya1,2 and Karthik Sukumar2

1 Cloud Computing and Distributed Systems (CLOUDS) Laboratory, Dept. of Computer Science and Software Engineering. The University of Melbourne, Parkville, VIC 3010, Australia2 Manjrasoft Pvt. Ltd., ICT Building, 111, Barry Street, Carlton, Melbourne, VIC 3053, Australia. {karthik, raj}@manjrasoft.com

Cloud computing is rapidly emerging as a new paradigm for delivering IT services as utlity-oriented services on subscription-basis. The rapid development of applications and their deployment in Cloud computing environments in effi cient manner is a complex task. In this article, we give a brief introduction to Cloud computing technology and Platform as a Service, we examine the offerings in this category, and provide the basis for helping readers to understand basic application platform opportunities in Cloud by technology’s such as Microsoft Azure, Sales Force, Google App, and Aneka for Cloud computing. We demonstrate that Manjrasoft Aneka is a Cloud Application Platform (CAP) leveraging these concepts and allowing an easy development of Cloud ready applications on a Private/Public/Hybrid Cloud. “Aneka CAP” offers facilities for quickly developing Cloud applications and a modular platform where additional services can be easily integrated to extend the system capabilities, thus being at pace with the rapidly evolution of Cloud computing.

1. Introduction Cloud computing is rapidly emerging

as a new paradigm for delivering computing as a utility [1]. It allows leasing of IT capabilities whether they are infrastructure, platform, or software applications as services on subscription-oriented services in a pay-as-you-go model. Its foundation is based on various developments in IT during the last thirty to forty years. As fresh ideas and technology advancement have made it all the more striking and appealing during the Internet age, the way consumers consume and technology enablers deliver solutions has evolved. With a trend towards Cloud based model, the power is shifted to consumers. They have access to more compute power and to new applications, at an alluring price, as well as they enjoy the advantages of a self-service and self-managed environment.

Cloud computing fosters elasticity and seamless scalability of IT resources that are offered to end users as a service through Internet medium. Cloud computing can help enterprises improve the creation and delivery of IT solutions by providing them to access services in a most cost effective and fl exible manner. A bird’s eye view of Cloud computing is shown in Figure 1.

Although Cloud computing has emerged mainly from the appearance of public computing utilities [2], various deployment models, with variations in physical location and distribution, have

Fig. 1 : A bird’s eye view of Cloud compu� ng.

ClientsPrivateCloud

Public Cloud

Other Cloud Services

Govt. Cloud Services

CloudManager

Karthik SukumarRajkumar Buyya

CSIC May 2011.indd 6 5/9/2011 2:36:39 PM


been adopted. In this sense, regardless of its service class, Clouds can be classifi ed as public, private, or hybrid depending on the model of deployment. A public Cloud is a Cloud made available in a pay-as-you-go manner to the general public. A private Cloud is a data center of an organization, not made available to the general public. A hybrid Cloud is a seamless use of public Cloud along with private Cloud when needed. In a typical public Cloud scenario, a third-party vendor delivers services such as computation, storage, networks, virtualization and applications to various customers. In a private Cloud environment, internal IT resources are used to serve their internal users and customers. Businesses are adopting public Cloud services to save capital expenditure and operational cost by leveraging Cloud’s elastic scalability and market oriented costing features. Nevertheless, public Cloud computing also raises concerns about data security, management, data transfer, performance, and level of control.

Cloud Computing started with a risk-free concept: let someone else take the ownership of setting up IT infrastructure and let end-users tap into it, paying only for what is been used. From this simple idea, a much more sophisticated, complex (and sometimes complicated) market started to grow. Today, businesses can buy computation resources, infrastructure plus platform or infrastructure plus applications. In the language of this market, the computation resources is frequently referred to as Infrastructure as a Service (IaaS), and the applications as Software as a Service (SaaS). In fact, use of the acronym appears ubiquitously from SaaS to PaaS (Platform as a Service) to XaaS (Anything as a Service). Key characteristics and vendors offering these Cloud services are highlighted in Fig.2.

What makes Cloud computing different from traditional IT approaches is the focus on service delivery and the consumer utilization model. In the background, service provider’s uses particular technologies, system architecture, design and industry best practices to provide and support the delivery of service-oriented, elastically scalable environment serving multiple customers. This helps end users to have more agile and fl exible service oriented architecture for their application and

Category Characteristics Cloud Providers

Software as a Service (SaaS)

Software as a service (SaaS) refers to applications delivered as cloud services where customers are provided with applications that are accessible anytime and from anywhere.

� Google Apps � Zoho Offi ce � Salesforce.com � Microsoft Offi ce

Live

� Force.com � Google App Engine � Microsoft Azure � Boomi � Manjrasoft Aneka

� Amazon EC2 � Go Grid � Rack Space � Sun Grid � VM Ware � Zen Server

Platform as a Service (PaaS)

Platform-as-a-Service (PaaS) refers to environment for application development with seamless Integration with Cloud for application hosting

Infrastructure as a Service (IaaS)

Infrastructure-as-a-Service (IaaS) refers to on-demand computing capacity from a service provider which is virtualized hardware and storage

Applications on SaaS

Application Dev Platform

Infrastructure Service

Automation Provisioning

Virtualization

Management Package

Network /Processing / Storage

services. In a conventional IT scenario, most software companies have procured different components of their application middleware infrastructure layer from various vendors, and brought together these tools into a corporate environment using system integration services and tools. On the other hand, in a Cloud computing scenario, this practice is quite rare. Platform-as-a-Service solutions provide environment and applications development platforms for seamlessly integrating Cloud computing into existing application, services, and infrastructure with a market-oriented approach.

2. Cloud Application Development Platforms

Application development, deployment and runtime management have always been reliant on development platforms such as Microsoft’s .NET, WebSphere, or JBoss, which have been deployed on-premise traditionally. In the Cloud-computing context, applications are generally deployed by Cloud providers to provide highly scalable and elastic services to as many end users as possible. Cloud computing infrastructure needs to support many users to access and utilize the same application services, with elastic allocation of resources. This has led to enhancement in development platform technologies and architectures to handle performance, security, resource allocation, application monitoring, billing, and fault tolerance.

There are several solutions available in the PaaS market, to mention a few: Google App Engine, Microsoft Windows Azure, Force.Com, and Manjrasoft Aneka. Google App Engine provides an extensible runtime environment for web based applications developed with Java or Python, which leverage huge Google IT infrastructure. Windows Azure provides a wide array of Windows based services for developing and deploying windows based applications on the Cloud. It makes use of the infrastructure provided by Microsoft to host these services and scale them seamlessly. Aneka provides a more fl exible model for developing distributed applications and provides integration with external Clouds such as Amazon EC2 and GoGrid. Aneka offers the possibility to select the most appropriate infrastructure deployment without being tied to any specifi c vendor–a virtual infrastructure, a private datacenter or a server –thus allowing enterprises to comfortably scale to the Cloud when needed.

2.1 Windows AzureThe Windows Azure Platform [3]

consists of SQL Azure and the .NET services. The .NET services comprises of Access Control services and .NET service bus. Windows Azure is a platform with shared multitenant hardware provided by Microsoft. Windows Azure application development mandates the use of SQL Azure for RDBMS functionality, because that is the only coexisting DBMS

Fig. 2 : Cloud service types, characteris� cs, and vendors.

CSIC May 2011.indd 7 5/9/2011 2:36:50 PM


functionality accessible in the same hardware context as the applications.

2.2 Google App EngineGoogle App Engine is offered

by Google Inc. Its key value is that developers can rapidly build small web based applications on their machine and deploy them on the Cloud. A notable thing is that Google App Engine provides developers with a simulated environment to build and test applications locally with any operating system or any system that runs a suitable version of Python and Java language environments. Google uses the Java Virtual Machine with Jetty Servlet engine and Java Data Objects.

2.3 Force.comForce.com is a development

and execution environment that is independent for Salesforce.com. Force.com is the best approach for Platform-as-a-Service (PaaS) for developing CRM based application and, with regards to the design of its platform and the runtime environment is based on the Java technology. The platform uses a proprietary programming language and environment called Apex code, which it has a reputation for simplicity in learning and rapid development and execution.

2.4 Manjrasoft AnekaAneka [4] is a distributed

application platform for developing Cloud applications. Distributed means that Aneka can seam together any number of Windows based physical or virtual desktops or servers into a network of interconnected nodes that act as a single logical “application execution layer.” The middleware is managed and monitored with advanced tools that allow monitoring applications’ performance and the system status in order to meet the Service Level Agreements (SLAs) made with the users. Aneka-based Clouds can be deployed on a variety of hardware and operating systems including several fl avors of the Windows and Linux operating system families. This fl exibility allows Aneka to virtually harness almost all the different types of infrastructure and runtime environment to serve application execution on demand.

3. Aneka Cloud Application Platform

Aneka [4] is a platform for developing resource-intensive and elastic

applications and their deployment on Clouds. It can harness a huge variety of physical and virtual resources, ranging from desktops, clusters, to virtual datacenters, to provide a single logical “application execution layer”. The key components of the platform are depicted in Figure 3, which gives an overall view of Aneka from its foundations to the applications and the end user services. The platform is based on an extensible Service Oriented Architecture (SOA), which makes the integration of new components, incremental development of new features, and infrastructure deployment and confi guration seamless tasks.

Middleware. The platform features a homogeneous distributed runtime environment for applications. Such environment is built by aggregating together physical and virtual nodes hosting the Aneka container. The container is lightweight layer that interfaces with the hosting environment and manages the services deployed on a node. Services constitute the core logic of Aneka Clouds and each container hosts three different classes of services:• Fabric Services. Fabric services implement the fundamental operations of the infrastructure of the Cloud. These services include: high-availability and failover for improved reliability, node membership and directory, resource provisioning, performance monitoring and hardware profi ling. • Foundation Services. Foundation services constitute the core functionalities of the Aneka middleware. They provide a basic set of capabilities that enhance application execution in the Cloud. These services provide the infrastructure with added value and are both of use for system administrators and developers. Within this category we can list: storage management, resource reservation, reporting, accounting, billing, services monitoring, and licensing. Services in this level operate across all the range of supported application models. • Application Programming Services. Application services deal directly with the execution of applications and are in charge of providing the appropriate runtime environment for each application model. At this level Aneka expresses its true potential in supporting different

application models and distributed programming patterns. Aneka provides support for the most known application programming patterns such as distributed threads, bag of tasks, and MapReduce. Application Development and Management. Aneka offers advanced features for developing and managing applications on the Cloud. The Software Development Kit (SDK) and the Management Kit are the two components exposing such capabilities. They provide means for interacting with the middleware and managing it with advanced user interfaces and bindings for applications. By using Aneka SDK, developers can quickly develop distributed applications, integrate the scaling capabilities of Aneka into existing applications, or implement new services to extend the potential of Aneka. The Management Kit allows deploying, managing, and tuning Aneka-based Clouds. By using a visual approach, it provides means to access and control every aspect of the middleware and also offers advanced features such as application reporting, accounting, billing, user management, and performance monitoring. The SDK and the Management Kit are the tools that enrich the user experience of developers and administrators respectively.

4. Cloud Application Programming Models and their Support in AnekaAneka Clouds aim to be a ubiquitous

environment serving any type of computing need of distributed applications. Therefore, they are expected to be fl exible enough to support several different models for developing applications: parameter sweep, concurrent, and data-intensive applications. In order to serve this purpose, Aneka provides engineers and developers with the concept of “Programming Model”, which is collection of abstractions and runtime support for expressing and developing distributed applications. The platform currently supports three different programming models. They are: Task, Thread, and MapReduce [5]. Moreover, its extensible architecture offers the freedom to plug other models into the existing infrastructure. By taking this approach Aneka is able to provide support for all the following types of distributed applications:

CSIC May 2011.indd 8 5/9/2011 2:36:51 PM


Fig. 3 : Aneka pla� orm architecture.

Parameter Sweep Applications: Task Programming. Task Programming provides developers with the ability of expressing applications as a collection of independent tasks. Each task can perform a different operation, or the same operation on different data, and can be executed in any order by the runtime environment. Task programming allows the parallelization of legacy applications on the Cloud. Application from various domains including scientifi c computing, fi nancial applications, media rendering and transcoding can be created using Task programming model. This model is the most popular in distributed computing and can be used as a starting point for implementing new models such as workfl ows with more complex requirements.

Concurrent Applications: Thread Programming. The main abstraction of

this model is the concept of thread which mimics the semantic of the common local thread but is executed remotely in a distributed environment. This model offers major control on the execution of the single components of an application but requires more management if compared with Task Programming, which is based on a “submit and forget” pattern. This model covers all the application scenarios of the Task programming and solves the additional challenges of providing a distributed runtime environment for local multi-threaded applications.

Data Intensive Applications: MapReduce. This model is an implementation of the MapReduce model, as proposed by Google, for .NET and integrated with Aneka. MapReduce has been designed to process a huge quantity of data by using simple operations that extracts useful information from a dataset

(the map function) and aggregates this information together (the reduce function). MapReduce can be a winning solution for data mining and analytic applications, bulk media processing, and content indexing. Aneka provides a solid support for the model and integrates it with all the other foundation services such as accounting and reporting, thus making this solution a competitive alternative within the same market segment.

5. High-Performance Cloud Applications

Aneka has been used in creating several interesting applications in domains such as life sciences, engineering, and creative media. Applications created using Aneka are able to run on enterprise or public Clouds without any change. The three case studies on the use of Aneka for building applications in engineering, geospatial, and life science domains are discussed below.

5.1 Manufacturing and EngineeringThe Manufacturing and Engineering

sectors include a wide range of market segments, from aerospace to automotive. Manufacturing organizations face a number of computing challenges as they seek to optimize their IT environments, including high infrastructure costs and complexity to poor visibility into capacity and utilization. Today’s design engineers need access to unrestrained, fl exible computing capacity on demand, so that design cycles can be as fast, cheap, and productive.

The GoFront group, a division of China Southern Railway, is responsible for designing the high speed electric locomotive, metro car, urban transportation vehicle and the motor train. The raw design of the prototypes requires high quality 3D images using Autodesk’s rendering software called Maya. By examining the 3D images, engineers identify problems in the original design and make the appropriate design improvements. However, such designs on a single four core served used to 3 days to render scenes with 2000 frames.

To reduce this time, GoFront has used Aneka and created an enterprise Cloud (see Fig. 4) within their company by utilizing networked PCs. They used Aneka Design Explorer, a tool for rapid creation of parameter sweep applications, in which the same program is executed many

Application Management Kit

Software Development Kit

Container

Programming Models

Foundation Services

Fabric Services

Infrastructure

Management Studio

APIs

Task Model

Membership Services

Dynamic Resource Provisioning Services

.NET @ Windows

Physical Machines/Virtual Machines

Mono @ Linux

Hardware Profi le Services

Reservation Services

Storage Services

License Services

Accounting Services

Thread ModelMap Reduce

ModelOtherModel

Security

Persistence

Design Explorer

SLA-Negotiation Web Services

Administration Portal

Management Web Services

Private Cloud

LAN networkData Center

Microsoft

IBM

Microsoft Google

CSIC May 2011.indd 9 5/9/2011 2:36:51 PM


times on different data items (in this case, executing the Maya software for rendering different images). A customized Design Explorer (called Maya GUI) has been implemented for Maya rendering. Maya GUI managed parameters, generated Aneka tasks, monitored submitted Aneka tasks and collected completed rendered images. The design image used to take three days to render (2000+ frames, each frame with more than fi ve different camera angles). Using only a 20 node Aneka Cloud, GoFront was able to reduce the rendering scenario from 3 days to 3 hours.

5.2 Geospatial Sciences and Technologies

Due to the continuous growth of GIS sciences and technologies, there have been even more geospatial and non-spatial data involved due to increase in number of data sources and advancement of data collection methodologies. Spatial analysis and Geo-computation are getting intricate and computationally demanding. The Department of Space, Government of India, adopted Aneka as the Cloud computing platform supporting the development of high performance

GIS applications [6]. Aneka enables a new approach to complex analyses of massive data and computationally intensive environments, and gives the opportunity to satisfy all the requirements of a high-performance and distributed GIS environment over the public, private and hybrid Clouds.

5.3 Health and Life ScienceWith the high volume and density of

data, along with the growing complexity of IT ecosystem and the pressures of competition and regulatory groups, life sciences organizations need IT infrastructure and management tools that can respond quickly to changing needs and, more importantly, enable rather than hamper the ability to innovate.

Aneka enables faster execution and massive data computation in life science R&D, clinical simulation, and business intelligence tools. It helps organizations to achieve greater levels of innovation in shorter timeframes while maximizing license utilization, increasing ROI, and realizing signifi cant savings over Cloud based technology. For its application in real time scenario, Jeeva, an Enterprise Cloud enabled portal for protein secondary

structure prediction, was developed based on Aneka. Research scientists use the portal to discover new prediction structures using parallel execution methods. The prediction took 20 minutes to complete when compared with the previous computational time of 8 hours. Also, Aneka has enabled implementation of personal health monitoring system aiding rehabilitation of stroke patients on public Cloud platform such as Amazon EC2.

5.4 IT Education and ResearchAs the IT fi eld is rapidly moving

towards Cloud Computing, software industry’s focus is shifting from developing applications for PCs to Data Centers and Clouds that enable millions of users to make use of software simultaneously. This is creating a huge demand for manpower with skills in this area. Educational and research organizations require a platform that can support (1) multiple models of application programming, (2) multiple types of Cloud deployments (private, public, or hybrid), and (3) extensible framework enabling educators/researchers to develop their own programming models and application

Fig. 4 : Rending images of locomo� ve design on GoFront’s private Cloud using Aneka.

Aneka Maya Renderer

Use private Aneka Cloud

LAN network(Running Maya Batch Mode on demand)

GoFront Private Aneka Cloud

Raw Locomotive Design Files (Using AutoDesk Maya)

Setup 2: Aneka Enterprise Cloud

Setup 1: Single Server

4 crores server

Single Server

7580

50

40

20

0

50

25

3.3810

6000 Frames

4000 Frames

2000 Frames

Aneka Cloud

Aneka utilizes idle desktops (30) to decrease task time

from days to hours

Time (in hrs)

Using Maya Graphical Mode

Directly

CSIC May 2011.indd 10 5/9/2011 2:36:53 PM


schedulers.Since Aneka allows one to build a

private Cloud with minimal investment by harnessing their existing IT resources (e.g., LAN-connected PCs), it has emerged as an excellent platform for teaching and research in Cloud computing area. A number of institutions in India such as MSRIT-Bangalore and C-DAC (Center for Development of Advanced Computing), Hyderabad, and other countries have used Aneka for setting up a Cloud Computing Lab and used it to offer practical exposure to their students (studying Cloud/Grid/High-Performance computing courses) in addition to building applications. Researchers in National Institute of Technology (NIT-Karnataka) have used Aneka for developing and evaluating new QoS-based resource provisioning and application scheduling algorithms. For case studies and to download Aneka platform, please visit http://www.manjrasoft.com/

6. Conclusions and Future Directions

The growing interest in Cloud computing has led to new approaches for allocating fi nancial resources and leveraging IT infrastructure and services. Cloud computing provides concrete opportunity for making a fl exible use of IT by turning it into a utility. Cloud adoption is becoming a standard practice in many business sectors to scale IT infrastructure on demand. Despite this, the development of elastic and scalable applications is a complex task. Cloud application development platforms offer huge cost savings by reducing the cost

of software engineering and enabling intelligent use of Cloud infrastructures. A wide range of applications scenarios from fi nancial services, to entertainment and media, or manufacturing and engineering, demonstrates how Cloud technology can help increasing technology effi ciency and adoption. PaaS technologies help organizations to harness their existing computing infrastructure and/or rent public Cloud infrastructure in a seamless manner. The true benefi ts of the Cloud application development will become apparent when developing and deploying application on solutions such as Aneka.

As the fi eld of Cloud computing is rapidly progressing, there exist many opportunities for researchers and industrial developers to explore further. Key open issues that needs further investigation include: (1) Software Licensing, (2) Seamless integration of private and Cloud resources, (3) Security, Privacy and Trust, (4) Cloud “Lock-In” worries and Interoperability, (5) Application Scalability Across Multiple Clouds, (6) Clouds Federation and Cooperative Sharing, (7) Global Cloud Exchange and Market Maker, (8) Dynamic Pricing of Cloud Services, (9) Dynamic Negotiation and SLA Management, (10) Energy Effi cient Resource Allocation and User QoS, (11), Power-Cost and CO

2

emission issues and seamless use of renewable and non-renewable electricity energy sources, and (12) Regulatory and Legal Issues.

AcknowledgementsAll members of CLOUDS Lab at the

University of Melbourne and Manjrasoft

have contributed towards various developments reported in this paper. In particular, we would like to thank Christian Vecchiola, Yi Wei, Xingchen Chu, Dileban Karunamoorthy, Suraj Pandey and Rodrigo Calheiros for their contributions towards the recent developments in Aneka and improving the paper. We thank Professor Geoffrey Fox whose seminar presentation in CLOUDS Lab has infl uenced on the content of this paper.

References1. R. Buyya, C. Yeo, S. Venugopal, J. Broberg,

and I. Brandic, Cloud Computing and Emerging IT Platforms: Vision, Hype, and Reality for Delivering Computing as the 5th Utility, Future Generation Computer Systems, 25(6):599-616, Elsevier, The Netherlands, June 2009.

2. R. Buyya, J. Broberg, and A. Goscinski (eds), Cloud Computing: Principles and Paradigms, Wiley Press, USA, Feb. 2011.

3. D. Chappell, Introducing the Windows Azure Platform, David Chappell & Associates, October 2010.

4. C. Vecchiola, X. Chu, and R. Buyya, Aneka: A Software Platform for .NET-based Cloud Computing, High Speed and Large Scale Scientifi c Computing, 267-295 pp., IOS Press, Amsterdam, Netherlands, 2009.

5. S Ghemawat and J Dean, MapReduce: Simplifi ed Data Processing on Large Clusters, Proceedings of the 6th Symposium on Operating System Design and Implementation (OSDI’04), San Francisco, CA, USA, 2004.

6. K . Raghavendra, A. Akilan, N. Ravi, K. P. Kumar, and G. Varadan, Satellite Data Product Generation Using Aneka Cloud, Research Demo at the 10th IEEE International Symposium on Cluster, Cloud, and Grid Computing (CCGrid 2010), Melbourne, Australia, 2010.

About the Authors

Dr. Rajkumar Buyya is Professor of Computer Science and Software Engineering; and Director of the Cloud Computing and Distributed Systems (CLOUDS) Laboratory at the University of Melbourne, Australia. He is also serving as the founding CEO of Manjrasoft., a spin-off company of the University, commercializing its innovations in Cloud Computing. He has authored and published over 300 research papers and four text books. Software technologies for Grid and Cloud computing developed under Dr. Buyya’s leadership have gained rapid acceptance and are in use at several academic institutions and commercial enterprises in 40 countries around the world. Dr. Buyya has led the establishment and development of key community activities, including serving as foundation Chair of the IEEE Technical Committee on Scalable Computing and fi ve IEEE/ACM conferences. These contributions and international research leadership of Dr. Buyya are recognized through the award of “2009 IEEE Medal for Excellence in Scalable Computing” from the IEEE Computer Society, USA. Manjrasoft’s Aneka Cloud technology developed under his leadership has received “2010 Asia Pacifi c Frost & Sullivan New Product Innovation Award”.

Karthik Sukumar is serving as a Director of Manjrasoft, Australia in the area of product management and business analysis. He draws from a strong background in Software as a Service having served as a product manager at India’s leading ISP, Sify Technology before joining Manjrasoft. He has performed and managed most of the roles of a modern software product organization and developed strategic partnership with leading organizations like IBM, Microsoft, Amazon, HP, VM Ware, and Citrix.

CSIC May 2011.indd 11 5/9/2011 2:36:58 PM


Technical Trends

1. Introduction Cloud computing is a model for

enabling convenient, on-demand network access to a shared pool of resources that can be rapidly provisioned and released with minimal management effort or service provider interaction [1]. By dynamically provisioning of resources enables cloud computing infrastructure to meet arbitrary varying resource and service requirements of cloud customer applications. The application requirements can be characterized by quality of service (QoS) requirements such as availability, security, reliability etc., as mentioned in the Service Level Agreement (SLA). SLA is a legal binding contract which states QoS guarantees that an execution environment (provider) agrees to provide its hosted application with. The Cloud Computing paradigm is composed of three service models namely Software as a Service (SaaS), Platform as a Service (PaaS), and Infrastructure as a Service (IaaS). SaaS model enables the customer to use the provider application hosted on the cloud infrastructure. In this model the customer does not have any control over the underlying Cloud infrastructure but has little control over application confi guration settings. PaaS model provides the capacity to the consumer to deploy the customer created applications onto the Cloud Infrastructure using the programming language or tools supported by the Cloud Provider. The Consumer does not manage the underlying cloud infrastructure such as network, storage

etc. but has control over the deployed applications. IaaS model allows the customer to provision processing, storage, networks and other important software where a customer can deploy software such as operating system and applications. The consumer does not have control over the underlying infrastructure but has control over operating systems, storage, deployed applications.

A cloud environment is characterized by multiple providers, each with its own service terms management systems, operating platforms, and security levels. The specialized expertise and value-addition of a cloud service broker will help IT managers fi nd the right cloud offering, deploy their application in the cloud and manage it properly. Cloud brokers negotiate the best deals and relationships between cloud consumers and cloud providers. They can use specialized tools to identify the most appropriate cloud resource and map the requirements of application to it. They can also be dynamic by automatically routing data, applications and infrastructure needs based on some QoS criteria like availability, reliability, latency, price etc.

Cloud broker services are mainly classifi ed into three categories viz. Service Intermediation, Service Aggregation, and Service Arbitrage. Service Intermediation broker provides a service to a consumer that enhances a given service by adding some value on top to increase some specifi c capability. Service Aggregation brokerage service combines and integrates

into one or more services and ensures data are modeled across all component services and movement and security of data between the service consumer and multiple providers. Service Arbitrage is similar to cloud service aggregation but services being aggregated are not fi xed. In addition these services provide fl exibility and opportunistic choices for the service aggregator.

2. BackgroundM.A. Salehi and Rajkkumar Buyya

[2] proposed a user level broker with two market oriented scheduling policies to increase the computational capacity of the local resources by hiring resources from an IaaS provider and optimized time and cost of an application. Recently, Yicho Yang et al [3] described a cloud infrastructure service frame work and introduced a service oriented broker to provide guarantee data transmission and uniform mechanism for making resources via broker to maintain certain level of services to users. In this paper we have proposed a framework namely Cloud Quality of Service management Strategy(C-QoSMS) which has to be included in the cloud service broker. The C-QoSMS component added in the cloud broker enables the customer to select a cloud provider based on the QoS criteria specifi ed in the SLA in minimum searching time. The searching time is minimized by using a concept called reduct [4][5] used in Rough Set theory [6][7][8]. The reduction of attributes is achieved by comparing equivalence relations

Quality of Service (QoS) is a broad term used to describe the overall experience a user or application will receive over a network. In our work a Cloud-QoS Management Strategy (C-QoSMS) is proposed to be added in resource broker of cloud environment. C-QoSMS allocates resources based on Service Level Agreement between users and providers for Infrastructure and a service (IaaS) cloud. A Soft computing technique is used to allocate the best provider to the cloud’s user with minimum searching time.

Quality of Service Design in Clouds

Praveen Ganghishetti and Rajeev Wankar*Dept. of Computer and Information Sciences, University of Hyderabad, Hyderabad, India{gpraveen86, rajeev.wankar}@gmail.com, (*corresponding author)

Praveen Ganghishetti

Rajeev Wankar

CSIC May 2011.indd 12 5/9/2011 2:36:59 PM


generated by sets of attributes. Attributes are removed so that the reduced set provides the same predictive capability of the decision feature as the original. A reduct is defi ned as a subset of minimal cardinality R

min of the conditional attribute

set C such that γR(D) = γC(D) [5].R = {X : X ⊆ C; γX (D) = γC(D)}

Rmin

= {X : X ∈ R; ∀ Y ∈ R; |X| ≤ |Y| }

3. System OverviewIn this section we present

Architecture of C-QoSMS framework and QoS parameters

3.1 Architecture of C-QoSMS Framework

In C-QoSMS framework the QoS can be classifi ed into three layers namely, Application layer QoS (ALQoS), Middleware Layer QoS (MLQoS) and Netware Layer Qos (NLQoS). ALQoS guarantees the requirements for the user and application when remote service is accessed. Elements of Application performance within the scope of ALQoS often include: Accessibility, the number of users accessing a service, Security, the authentication and authorization mechanisms, and availability i.e. how much time a service is available. MLQoS consists of the response (access time) of the required service that will be the same as user’s requirement, even when the service is shared among multiple users. It is relevant to the type of applications which needs an interaction between user and the application such as visualization applications. Elements of Middleware performance within the scope of QoS often include the frequency of the CPU, Secondary Memory Storage, Main Memory Storage, Cluster and GPU. NLQoS is a broad collection of networking technologies and techniques. The goal of NLQoS is to provide a network to deliver predictable results. Elements of network performance within the scope of QoS often include network availability (uptime), bandwidth (throughput), latency (delay), and error rate.

In the present work, C-QoSMS frame work is employed in the cloud broker to discover the best service provider among all the available service providers. The term “best” means, the elements of all QoS layers in the provider match the requirements of the cloud user/ application.

Various parameters namely (1) Availability, (2) Security, (3) Processor frequency, (4) Main Memory Storage, (5) Secondary Memory Storage, (6) GPU, (7) I/O performance and (8) Network Round Trip Time (RTT) of the three QoS layers have been considered in this model. While all other parameters mentioned in the SLA are assumed to be met by the provider. 3.2 Modules of C-QoSMS:

The C-QoSMS framework has been divided into four modules.3.2.1 Cloud Registry (CR):

It is a registry service used to record service capability and QoS provisions for different cloud managers by different service providers. Here the QoS parameters recorded would be all elements of the three QoS layers.

3.2.2 Cloud Manager (CM):Service provider consists of CM

that is used to host the application of the client. Eucalyptus [10] environment may be considered as the service provider for service and execution environment with QoS specifi cation for Infrastructure as a Service (IAAS).

3.2.3 Network Resource Manager (NRM):

NRM uses the concept called “Bandwidth Broker” which is responsible for managing the communication between the user and the cloud manager.

3.2.4 Application QoS Manager (AQoSM):

AQoSM has a specifi c strategy using CQoSMS algorithm to fi nd service providers adhering QoS requirements as specifi ed for a service in an SLA. When a service provider has been selected, the AQoSM coordinates with the service providers cloud manager for subsequent service execution.

In general cloud providers publish their service along with all types of QoS parameters in the cloud registry. Then C-QoSMS algorithm will be invoked. The steps of the C-QoSMS algorithm are given below:

3.3 C-QoSMS AlgorithmStep-1: A cloud user/application submits a request to the cloud AQoSM, to get a specifi c service (specifi cations of infrastructural service including its QoS parameters). Step-2: Application QoS Manager of

the cloud broker will get the available list of providers with QoS specifi cations required for the requested service from the cloud registry. Step-3: A list of providers with their attributes is sent to the optimization strategy. The main objective of optimization strategy is to select the best provider from list of providers. In our model the reduct concept of Rough Set theory is used as an optimization strategy.

The following steps give the selection part of the algorithm which is used in the proposed model:Step-4: Constructing the information table containing list of providers with all related attributes (ALQoS and MLQoS parameters) from the cloud registry.Step-5: Clustering is done on the providers (objects) considering ALQoS and MLQoS parameters. Step-6: Clusters labels are used as a decision attribute to form the Decision table.Step-7: Apply reduct algorithm to get all the possible reducts from the decision table.Step-8: Select one reduct having all preferable attributes by the user/application or by the requested service.Step-9: We now get a reduced information table with the reduced set of attributes and distinct clusters. Euclidean distance Equation is used between the selected service QoS parameters to all the distinct clusters of providers.Step-10: The cluster which scored the least distance is selected. Now, AQoSM interacts to monitor the network links between the selected reduct of providers and the broker.Step-11: NRM after monitoring the networks gets all the RTT’s (NLQoS parameter) between the broker and all the providers of the selected cluster.Step-12: The provider with the least RTT is picked by NRM to reserve a link between the user and the selected provider.

3.4 Interaction with NRM:NRM is used to monitor the network

links between the users and the providers. AQoSM interacts with NRM at two different occasions:1. AQoSM after applying reducts gets

minimized set of providers. Now, AQoSM interacts with NRM to monitor provider’s network links which closely matches with Network

CSIC May 2011.indd 13 5/9/2011 2:37:01 PM


QoS parameters as specifi ed in the SLA. NRM then gives back provider’s RTT to AQoSM.

2. AQoSM interacts with NRM for provisioning of network resources monitoring and supporting admission control once an appropriate

service has been discovered by the strategy used by AQoSM. The NRM determines a communicated path between user and the destination. Then the Network path gets reserved using protocols like RSVP as per the Network QoS parameters specifi ed in

the SLA for the service.There is no single universally

accepted standardization available in the literature. In our work we used Amazon’s various MLQoS Parameters that are given in the Tables1-5 [9] while ALQoS parameters are given in Tables 1-8.

Table 1 : Main memory range and Standardized valueMain memory range (GB) Standardized value

Below 1.7 1

1.7 - 6.99 2

7 - 7.4 3

7.5 - 14.9 4

15 - 17 5

17.1 - 21.9 6

22 - 22.9 7

23 -34.1 8

34.2 – 68.3 9

Above 68.3 10

Table 2 : Secondary memory range and Standardized valueSecondary Memory Range (GB) Standardized value

EBS Storage 1

160 - 349.9 2

350 - 419.9 3

420 - 849.9 4

850 - 1689.9 5

Above 1690 6

Table 3 : Processor Frequency Range and Standardized valueProcessor Frequency Range Standardized value

2 EC2 1

1 VC with 1 EC2 2

2 VC with 2 EC2 3

2 VC with 2.5 EC2 4

2 VC with 3.25 EC2 5

4 VC with 2 EC2 6


8 VC with 2.5 EC2 8


33.5 EC2 10

EC2: One EC2 Compute Unit provides the equivalent CPU capacity of a 1.0-1.2 GHz 2007 Opteron or 2007 Xeon processor

VC: virtual core

Table 4: Input/output and Standardized value

Input/output Range Standardized value

Below 10 Gigabit Ethernet 1

10 Gigabit Ethernet and above 2

Input/output performance refers to the transfer rate between the

main memory and secondary storage.

Table 5 :GPU and Standardized value

GPU Standardized value

2 x NVIDIA Tesla “Fermi” M2050 GPUs

1

Table 6 :Service Availability and Standardized value

Service Availability Standardized value

Below 99% 1

99% and above 2

Table 7 :Security Mechanism and Standardized value

Security Mechanism Standardized value

Simple 1

Complex 2

In simple security mechanism only provider gets authenticated where as in Complex security mechanism double authentication takes place where both client and provider gets authenticated.

3.5 Standardization of QoS parameters of various Applications Selected by the client:

3.5.1 Infrastructural Service:C-QoSMS framework meets the requirements of the

customer of all the three cloud service models. However, we illustrate for Infrastructural Service in this work. According to Amazon [9] the applications opted by client can be broadly classifi ed into six categories. They are:1. General Purpose Applications: The memory to CPU ratios of

these applications is given below:

Category Main Memory Processor Storage

(i) 2 2 2

(ii) 4 3 5

(iii) 5 6 6

2. Database and Memory Caching Applications: The resources required for these applications should have larger memory sizes. The QoS parameters can fall into one of these categories.


(i) 6 5 4

(ii) 9 7 5

(iii) 10 9 6

CSIC May 2011.indd 14 5/9/2011 2:37:01 PM


3. Website Based Applications: These applications require a small amount of consistent CPU resources and can increase the CPU capacity when additional cycles are required.


(i) 1 1 1

4. Compute Intensive Applications: Application that requires more CPU resources than main memory come under this category.


(i) 2 4 3

(ii) 3 8 6

5. High-Performance Compute Applications: The applications which required large amount of CPU coupled with increased network performance come under this category.

Category Main Memory Processor Storage I/O

(i) 8 10 7 2

6. Rendering and Media Processing Application: These

applications should have general purpose graphics processing unit with proportion to high CPU and increased network performance.

Category Main Memory Processor Storage I/O GPU

(i) 7 10 7 2 1

One of the above applications that have been selected by the client will be processed against the list of providers that the cloud broker has and maps a cloud provider which closely matches the QoS requirements.

Conclusion:To meet the requirements of both cloud users and service

providers, effective and effi cient resource broker is proposed with new C-QoSM management. Rough Set Theory is used to minimize number of attributes and minimize Searching space. The reduct algorithm is simulated in RSES tool and received reasonably good results.

Acknowledgement:Authors wish to register their sincere thanks to Ms. Rafah M. Almuttairi and Prof. C. Raghvendra Rao for their valuable academic advice for carrying out this research work.

References:1. http://csrc.nist.gov/groups/SNS/cloud-

computing/2. Mohsein Amini Salehi, Rajkumar Buyya,

“Adapting Market-Oriented Scheduling policies for Cloud Computing”, Algorithms and Architecture for Parallel Processing Lecture Notes in Computer Science, 2010, Volume 6081/2010, pp. 351-362.

3. Yichao Yang, Yanbo Zhou, Lei Liang, Dan He, Zhili Sun, “A Service-Oriented Broker for Bulk Data Transfer in Cloud Computing”, 2010 Ninth International Conference on Grid and Cloud Computing November 01-05, ISBN: 978-0-7695-4313-0.

4. Chuanjian YANG, Hao GE2, Guangshun

YAO, Lisheng MA, “Quick Complete Attribute Reduction Algorithm”, Sixth International Conference on Fuzzy Systems and Knowledge Discovery IEEE Computer Society Washington, DC, USA 2009 Volume 04( ISBN: 978-0-7695-3735-1).

5. K. Thangavel, Qiang Shen, A. Pethalakshmi, “Application of Clustering for Feature Selection Based on Rough Set Theory Approach”, The International Journal of Artifi cial Intelligence and Machine Learning (AIML) Journal, Volume (6), Issue (1), January, 2006.

6. Pawlak, Z., “Rough sets”, International Journal of Computer and Information Sciences. Vol. 11, pp: 341–356,198.

7. Pawlak, Z., “Rough Classifi cation”, Int. J. Man-Machine Studies. Vol. 20, pp: 469–483, 1984.

8. Polkowski, L. and Skowron, A. (Eds.), “Rough Sets in Knowledge Discovery”, Physica-Verlag 1(2) 1998.

9. http://aws.amazon.com/ec2/10. Daniel Nurmi, Rich Wolski, Chris

Grzegorczyk, Graziano Obertelli, Sunil Soman, Lamia Youseff, Dmitrii Zagorodnov, “The Eucalyptus Open-Source Cloud-Computing System”, Proceeding CCGRID ‘09 Proceedings of the 2009 9th IEEE/ACM International Symposium on Cluster Computing and the Grid IEEE Computer Society Washington, DC, USA.

About the Authors

Mr. Praveen Ganghishetti is pursuing his M.Tech. in University of Hyderabad, Hyderabad. He did his B.Tech in Computer Science from JNTU, Hyderabad in 2007. Praveen has about 3 years of industrial experience in Software Development in a reputed organization. His research interests include Cloud computing and Data mining.

Dr. Rajeev Wankar is working as a Reader (Associate Professor) in the Department of Computer and Information Sciences at University of Hyderabad. He earned Ph.D. in Computer Science from the Department of Computer Science, Devi Ahilya University Indore. In 1998, the German Academic Exchange Service awarded him “Sandwich Model” fellowship. He was working in the Institut für Informatik, Freie Universität, Berlin and had collaboration with Scientists of Konrad Zuse Institut für Informationstechnik (ZIB), a Supercomputing Laboratory in Berlin, for almost two years. Currently he is working (research/teaching) in the area of Parallel Computing, Distributed Shared Memory Computing, Grid Computing and Multi Core Computing. He is actively participating in an International Geo-Grid activity known as GEON with San Diego Supercomputing Centre, University of California, San Diego from Indian side and an Associate Faculty in the University Centre of Earth and Space Sciences (UoH). He served as a program committee member in many conferences such as HiPC-07, TEAA, ICDCIT, TENCON etc. He served as the Guest Editor of IJCSA’s Special issue on Grid and Parallel Systems.

CSIC May 2011.indd 15 5/9/2011 2:37:01 PM


Research Front

Resource Management on Clouds: Handling Uncertainties in Parameters and Policies

1. IntroductionCloud computing that is based

on resources acquired on demand is generating a great deal of interest among service providers and consumers as well as researchers and system builders. The rise in interest is refl ected in a signifi cant and continuous annual growth in the virtualization market. Both Meryl Lynch and Gartner have predicted a multi-billion dollar market for the cloud computing industry. As pointed out in [6], the annual world-wide enterprise IT spending on cloud computing is expected to increase from 4% of the total IT budget in 2008 to 9% by 2012. The primary reasons for the popularity of this new distributed system infrastructure include the following.

� Low IT investment: With cloud computing the service consumer does not need to purchase resources. The “pay as you go” feature of the cloud allows users to acquire resources on demand for an hourly resource rental fee for example. This is of great value for small start up companies as well as larger enterprises. The start up companies do not need a heavy investment on setting up an IT infrastructure whereas the cost of IT management and upgrading is greatly reduced for the larger enterprises.

� Elasticity: A cloud lets the computing demand of a service consumer grow and shrink dynamically in accordance with its current resource requirement. This provides a signifi cant benefi t for handling temporary increases in resource usage for the service consumer.

� Green Computing: By consolidating the IT operations of multiple consumers at a single data centre, an effective resource sharing is achieved leading to a reduction in power consumed by the computing and cooling equipments.Various types of clouds are in

operation today. These include public clouds such as the one provided by the Amazon Elastic Compute Cloud (EC2) that comprise shared resources that can be used by the public consumers at large. A private cloud on the other hand is accessible to the members of a given group only. An enterprise cloud that serves the employees of a given company and a research and engineering cloud that unifi es resources located in multiple institutions are two variants of the private cloud. Irrespective of the type of a cloud

performing resource management effectively is an important concern. A recent survey shows that security and performance are two top priorities for cloud service consumers [5]. An effective resource management strategy is crucial for harnessing the power of the underlying distributed hardware and achieving a high system performance. As in the case of grids [4], a predecessor of cloud, quality of service (QoS) remains an important issue. Service Level Agreement (SLA), an important characteristic of clouds [1] often requires the handling of an Advance Reservation (AR) request that is characterized by a deadline.

Although cloud computing is an effective and economic solution for distributed computing it poses a number of challenges. These include:

� Lack of Control: By delegating the

Shikharesh MajumdarReal Time and Distributed Systems Research Centre, Department of Systems and Computer Engineering,Carleton University, Ottawa, CANADA. Email: [email protected]

Appropriate management of resources by cloud middleware is crucial for effectively utilizing the underlying distributed resource infrastructure. Existing literature on resource management on grids and clouds describes techniques that are based on a detailed knowledge of local resource management policies as well as user estimates of resource demands for their requests. It is often impractical to assume that such a detailed a priory knowledge of management policies for all the resources will be available to resource brokers in a large, dynamic and heterogeneous cloud environment. Moreover, user estimates of resource demands are often error prone. Performing effective resource management by adequately handling such uncertainties associated with the knowledge of local resource management policies and user estimated resource demands is discussed.

CSIC May 2011.indd 16 5/9/2011 2:37:01 PM


IT management to a third party the service consumer loses the ability to directly control and monitor the execution of the submitted workload.

� Security: Trusting a third party to store an enterprise’s confi dential data is a source of potential concern and needs to be effectively addressed by the cloud service provider.

� Inter-operability among multiple service providers: To avoid vendor lock-in it is desirable to be able to run the same virtualized application on clouds provided by multiple vendors. Appropriate standards need to be in place for addressing this concern.

� Resource Management: Management of computing, storage and other resources in a distributed system is recognized as a “hard” problem. The dynamic nature of the resource pool comprising a large number of heterogeneous resources increases the complexity of resource management on a cloud signifi cantly. The handling of a service level agreement adds to the challenges of devising effective resource management techniques. Various uncertainties that are associated with satisfying SLAs on an environment with a large and dynamic resource pool need to be handled and are discussed in more detail in the following sections. The rest of the paper focuses

on resource management on clouds, describing further details of the underlying challenges and providing a high level outline of a number of techniques for addressing these concerns.

2. Resource Management on Clouds

Managing resources on clouds require effectively performing two operations: matchmaking and scheduling. Matchmaking is the process of allocating jobs associated with user requests to resources selected from the available resource pool. Scheduling refers to determining the order in which jobs mapped to a specifi c resource are to be executed. Effective matchmaking and scheduling are crucial for utilizing the power of the underlying resource pool and achieving high system performance that in turn leads to user satisfaction and enhanced revenue for the service

Fig. 1 : A Resource Management Framework (from [2])

provider. Because of the similarity of a cloud with a utility grid, many research results existing in the domain of resource management on grids can be extended to the cloud domain.

A cloud workload is typically characterized by two types of requests: Advance Reservation and On Demand (OD) requests. With an Advance reservation request a service requester engages in an SLA with the service provider. An AR is characterized by an earliest start time, an execution time for the job associated with the request and a deadline by which the job must be completed. An OD on the other hand does not have a deadline and needs to be completed on a best effort basis. An example framework for resource management on grids is presented in Figure 1. A detailed description of the framework in provided in [2]. This framework can be easily extended to the cloud domain. Only some of the components are described in this paper. The reader is directed to [2] for further details.

Requests are submitted to the Multi-Resource Coordinator (MMC) and the Matchmaker maps each request to

an appropriate resource. The Resource Liaison and Controller (RLC) performs admission control for each resource and ensures that an AR allocated to the resource is able to meet it deadline. The Resource Manager within RLC negotiates access to a resource with the help of its Local Resource Manager. The Match Advisor determines a degree of fi t between an AR and a resource and passes on this information to MMC for performing an effective matchmaking. The Scheduler performs the scheduling of each request in such a way that its deadline (if any) is met while a high resource utilization is achieved. An example scheduler that uses an algorithm called Scaling through Subset Scheduling is described in [3]. Some of the other components associated with request scheduling are discussed in Section 3.1.

3. Uncertainties Affecting Resource Management

Uncertainties associated with job execution times and the strategy used by the Local Resource Manager for scheduling signifi cantly increases the complexity of resource management.

CSIC May 2011.indd 17 5/9/2011 2:37:05 PM


Two types of uncertainties that increase diffi culty in scheduling and matchmaking are discussed next.

3.1 Error Associated with Estimation of Job Execution Times

In order to handle ARs a cloud resource manager needs information regarding the execution times of jobs. Estimating the execution time for a job is a hard task for a user and errors are made very often (see [9] for example). Existing literature shows that the execution time estimates for jobs submitted by users can be grossly over estimated. Jobs may run for a smaller time in comparison to their stipulated execution times also when an abnormal job termination occurs because of a run time error for example. These give rise to resource idle times and lead to a serious degradation in system performance because jobs that could have used the resource during these idle time periods might have been turned away by the matchmaker that expected the resource to be busy exec

cocsi-india.org.in/communications/csic may 2011.pdf · 2019. 1. 15. · csi report : csi discover...

Documents