1. I n t e r n a t i o n a lT e l e c o m m u n i c a t i o n U n i o n Distributed Computing: Utilities, Grids &CloudsITU-T Technology Watch Report 92009Terms such as Cloud Computing have gained a lot of attention, as they are usedto describe emerging paradigms for the management of information and computingresources. This report describes the advent of new forms of distributed computing,notably grid and cloud computing, the applications that they enable, and theirpotential impact on future standardization. Telecommunication Standardization Policy Division ITU Telecommunication Standardization Sector

2. ITU-T Technology Watch ReportsITU-T Technology Watch Reports are intended to provide an up-to-date assessment of promising new technologies in a language that is accessible to non-specialists, with a view to: Identifying candidate technologies for standardization work within ITU. Assessing their implications for ITU Membership, especially developing countries.Other reports in the series include:#1Intelligent Transport System and CALM#2Telepresence: High-Performance Video-Conferencing#3ICTs and Climate Change#4Ubiquitous Sensor Networks#5Remote Collaboration Tools#6Technical Aspects of Lawful Interception#7NGNs and Energy Efficiency#8Intelligent Transport Systems Acknowledgements This report was prepared by Martin Adolph. It has benefited from contributions and comments from Ewan Sutherland and Arthur Levin. The opinions expressed in this report are those of the authors and do not necessarily reflect the views of the International Telecommunication Union or its membership. This report, along with previous Technology Watch Reports, can be found at www.itu.int/ITU-T/techwatch. Your comments on this report are welcome, please send them to tsbtechwatch@itu.int or join the Technology Watch Correspondence Group, which provides a platform to share views, ideas and requirements on new/emerging technologies. The Technology Watch function is managed by the ITU-T Standardization Policy Division (SPD). ITU 2009All rights reserved. No part of this publication may be reproduced, by any means whatsoever, without the prior written permission of ITU. 3. ITU-T Technology Watch Reports Distributed Computing:Utilities, Grids & CloudsThe spread of high-speed broadbandextend the geographical reachofnetworks in developed countries, thedistributed computing, as the client-servercontinual increase in computing power, andrelationship can extend across borders andthe growth of the Internet have changed continents.the way in which society managesA number of new paradigms and termsinformation and information services.related to distributed computing have beenGeographically distributed resources, suchintroduced, promising to deliver IT as aas storage devices, data sources, and service. While experts disagree on thesupercomputers, are interconnected andprecise boundaries between these newcan be exploited by users around the worldcomputing models, the following tableas single, unified resource. To a growing provides aroughtaxonomy.extent, repetitive or resource-intensive ITtasks can be outsourced to serviceNewNew ServicesNew orproviders, which execute the task and often Computingenhancedprovide the results at a lower cost. A newParadigmsFeaturesparadigm is emerging in which computing isoffered as a utility by third parties whereby Cloud Software as aUbiquitouscomputingService (SaaS)accessthe user is billed only for consumption.This service-orientedapproachfrom EdgeInfrastructure Reliabilityorganizations offering a large portfolio of computingas a Service Scalabilityservices can be scalable and flexible. (IaaS)Grid VirtualizationThis report describes the advent of new computingPlatform as a Service (PaaS)Exchangeabil-forms of distributed computing, notably gridUtilityand cloud computing, the applications thatcomputingService- ity / Location Orientedindependencethey enable, and their potential impact onfuturestandardization. The idea of ArchitectureCost-distributing resources within computer (SOA) effectivenessnetworks is not new. It dates back to It is difficult to draw lines between theseremote job entry on mainframe computers paradigms: Some commentators say thatand the initial use of data entry terminals.grid, utility and cloud computing refer toThiswas expandedfirst withthe same thing; others believe there areminicomputers,thenwithpersonalonly subtle distinctions among them, whilecomputers (PCs) and two-tier client-serverothers would claim they refer to completelyarchitecture. While the PC offered more different phenomenon.ii There are no clearautonomy on the desktop, the trend is or standard definitions, and it is likely thatmoving back to client-server architecture vendor A describes the feature set of itswith additional tiers, but now the server iscloud solution differently than vendor B.not in-house. Thenewparadigms are sometimesNot only improvements in computer analogized to the electric power grid, whichcomponent technology butalso in provides universal access to electricity andcommunication protocols paved the way for has had a dramatic impact on social anddistributed computing. Networks based onindustrial development.iii Electric powerSystems Network Architecture (SNA), grids are spread over large geographicalcreated by IBM in 1974, and on ITU-Tsregions, but form a single entity, providingX.25, approved in March 1976i, enabledpower to billions of devices and customers,large-scale publicand privatedata in a relatively low-cost and reliablenetworks. These were gradually replaced byfashion.iv Although owned and operated bymore efficient or less complex protocols, different organizationsat differentnotablyTCP/IP. Broadband networks geographical locations, the components ofDistributed Computing: Utilities, Grids & Clouds (March 2009) 1 4. ITU-T Technology Watch Reports Figure 1: Stack of a distributed systemClients (e.g., web browser, and other locally installed software, devices)Middleware services (e.g., for load balancing, scheduling, billing)Resource entity 1 Resource entity 2Resource entity 3Resource entity n (e.g., application(e.g., virtual(e.g., database, server) system) storage)Resource interconnecterShared resources grids appear highly heterogeneous in theirVirtual resources, which can be physical characteristics. Its users onlyexchanged and are independent from its rarely know about the details of operation, physical location; like virtual memory or the location of the resources they areo Operating systems using.o Software and licenses In general terms, a distributed system is is o Tasks and applications a collection of independent computers thato Services appears to its users as a single coherent Figure 1 outlines a possible composition of system (Andrew S. Tanenbaum)v. A a distributed system. Similar system stacks second description of distributed systems have been described, e.g., specifically for by Leslie Lamport points out the importance cloudsviii and gridsix, and in a simplified of considering aspects such as reliability, stack with three layersx: application layer, fault tolerance and security when going mediator (=resource interconnecter), and distributed: You know you have a connectivity layer (=shared resources). distributed system when the crash of a computer youve never heard of stops youThe client layer is used to display from getting any work done.viinformation, receive user input, and to communicate with the other layers. A web Even without a clear definition for each of browser, a Matlab computing client or an the distributed paradigms: clouds and grids Oracle database client suggest some of the have been hailed by some as a trillion dollar applications that can be addressed. business opportunity.vii A transparent and network-independent Shared resourcesmiddleware layer plays a mediating role: it The main goal of a distributed computingconnects clients with requested and system is to connect users and IT resources provisioned resources, balances peak loads in a transparent, open, cost-effective, between multiple resources and customers, reliable and scalable way.regulates the access to limited resources (such as processingtimeon a The resources that can be shared in grids,supercomputer), monitors all activities, clouds and other distributed computinggathers statistics, which can later be used systems include:for billing and system management. ThePhysical resourcesmiddleware has to be reliable and always o Computational power available. It provides interfaces to over- o Storage devices and underlying layers, which can be used by programmers to shape the system o Communication capacity according to their needs. These interfaces enable the system to be scalable, extensible, and to handle peak loads, for 5. ITU-T Technology Watch ReportsBox 1: Amazon.com holiday sales 2002-2008 2002 20032004 2005 2006 20072008 Number of items1.7m 2.1m2.8m 3.6m 4m 5.4m6.3m ordered on peak day Average number of20 2432 41 46 62.572.9 items ordered per second on peak day Amazon.com, one of the worlds largest online retailers, announced that 6.3 millionitems were ordered on the peak day of the holiday season on 15 December 2008 amultiple of the items sold on an ordinary business day. This is 72.9 items per secondon average.Source: Amazon.com press releases, 2002-2008instance during the holiday season (see Box infrastructure in the world. This brings1). together more than 10,000 users in 140institutions (300 sites in 50 countries) toDifferent resources can be geographicallyproduce a reliable and scalable computingdispersed or hosted in the same dataresource available to the European andcenter.Furthermore, they can beglobal research community.xii High-energyinterconnected. Regardless of thephysics (HEP) is one of the pilot applicationarchitecture of the resources, they appeardomains in EGEE, and is the largest user ofto the user/client as one entity. Resourcesthe grid infrastructure. The four Largecan be formed into virtual organizations,Hadron Collider (LHC) experiments atwhich again can make use of otherCERNxiii, Europes central organization forresources.nuclear research, have a production whichProvided that the service meets the involves more than 150,000 daily jobs senttechnical specifications defined in a Service to the EGEE infrastructure and generatesLevel Agreement, for some users the hundreds of terabytes of data per year.location of the data is not an issue. This is done in collaboration with the OpenHowever, the users of distributed systems Science Grid (OSGxiv) project in the USAneed to consider legal aspects, questions ofand the Nordic Data Grid Facility (NDGFxv).liabilityand datasecurity, beforeThe CERN grid is also used to supportoutsourcing data and processes. Theseresearch communities outside the field ofissues are addressed later in this report.HEP. In 2006, the ITU-R Regional RadioGrid computingConference (RRC06xvi) established a newfrequency plan for the introduction of digitalGrid computing enables the sharing,broadcasting in the VHF (174-230 MHz) andselection, and aggregation by users of aUHF (470-862 MHz) bands. The complexwide variety of geographically distributedcalculations involved required non-trivialresources owned by different organizationsdependable computing capability. The tightand is well-suited for solving IT resource-schedule at the RRC06 imposed veryintensive problems in science, engineeringstringent time constraints for performing aand commerce.full set of calculations (less than 12 hoursGrids are very large-scale virtualized, for an estimate of 1000 CPU/hours on a 3distributed computing systems. They cover GHz PC).multiple administrative domains and enableThe ITU-R developed and deployed a client-virtual organizations.xi Such organizationsserver distributed system consisting of 100can share their resources collectively tohigh speed (3.6 GHz) hyper-thread PCs,create an even larger grid.capable of running 200 parallel jobs. ToFor instance, 80,000 CPU cores are shared complement the local cluster and to providewithin EGEE (Enabling Grids for E-sciencE), additional flexibility and reliability to theone of the largest multi-disciplinary gridplanning system it agreed with CERN to useDistributed Computing: Utilities, Grids & Clouds (March 2009) 3 6. ITU-T Technology Watch Reports Box 2: Folding@home: What is protein folding and how is folding linked to disease? Proteins are biologys workhorses, its nanomachines. Before proteins can carry out these important functions, they assemble themselves, or fold. The process of protein folding, while critical and fundamental to virtually all of biology, in many ways remains a mystery. Moreover, when proteins do not fold correctly (i.e. misfold), there can be serious consequences, including many well known diseases, such as Alzheimers, Mad Cow (BSE/CJD), Huntingtons, Parkinsons, and many cancers. Folding@home uses distributed computing to simulate problems millions of times more challenging than previously achieved, by interconnecting idle computer resources of individuals from throughout the world, represented as red dots in the Figure above (May 2008). More than 400,000 CPUs are active, corresponding to a performance of 4.5 PFLOPS. Source: http://folding.stanford.edu/ resources from the EGEE grid infrastructure the monthly compatibility analysis of HF (located at CERN and other institutions inbroadcasting schedules at nighttime. Germany, Russia, Italy, France and Spain). The resources of hundreds and thousands UNOSATxvii is a humanitarian initiative PCs are organized with the help of delivering satellite solutions to relief andmiddleware systems. The Berkeley Open development organizations within andInfrastructure for Network Computing outside the UN system for crisis response,(BOINCxxiii) isthe most widely-used early recovery and vulnerability reduction. middleware in volunteer computing made UNOSAT uses the grid to convert available to researchers and their projects. uncompressed satellite images into Grid technology has emerged from the JPEG2000 ECWxviii files. UNOSAT has scientific and academic communities and already been involved in a number of joint entered the commercial world. For instance, activities with ITU, particularly in providing the worlds largest company and banking satellite imagery for humanitarian workxix. groupxxiv HSBC uses a grid with more than In volunteer computing, individuals donate3,500 CPUs operating in its data centers in unused or idle resources of their computers four countries to carry out derivative to distributed computing projects such as trades, which rely on making numerous SETI@homexx, Folding@homexxi (see Box 2)calculations based on future events, and and LHC@homexxii. A similar mechanism risk analysis, which also looks to the future, has also been implemented by ITU-Rcalculating risks based on available utilizing idle PCs of ITUs staff to carry outinformationxxv. The German shipyard FSGxxvi useshighperformancecomputing 7. ITU-T Technology Watch Reportsresources to solve complex and CPU- commerce services, multiple other isolatedintensive calculations to create individual computing environments used by itsship designs in a short time. On-demand customers. These virtual machines areaccess to resources, which are not availablesoftware implementations of reallocally or which areonly needed computers that canbecustomizedtemporarily, reduces cost of ownership andaccording to the customers needs:reduces technical and financial risks in theprocessing power,storagecapacity,ship design. By increasing the availability ofoperatingsystem (e.g., Linux, MScomputing resources and helping toWindows), software, etc.integrate data, grid computing enablesWith theincreasing availability oforganizations to address problems thatbroadband networks in many countries,were previously too large or too complexsome computer utility providers do notfor them to handle alone. Other commercialnecessarily need to be geographicallyapplications of grid computing can be founddistributed or in close proximity to clients:in logistics, engineering, pharmaceuticalsproviders tend to build their data centers inand the ICT sector.xxviiareas with the lowest costs, e.g., forUtility computing electricity, real estate, etc. and with accessto renewable energy (e.g. hydroelectric).The shift from using grids for non- However, in many cases it proves useful tocommercial scientific applications to using employ data centers close to the customer,them in processing-intensive commercial for example to ensure low rates of latencyapplications led to also using distributedand packet loss in content deliverysystems for less challenging and resource-applications. For example, content deliverydemanding tasks.providers such as Akamaixxix or LimelightThe concept of utility computing is simple: Networksxxx built their networks of datarather than operating servers in-house, centers around the globe, and interconnectorganizations subscribe to an externalthem with high-speed fiber-opticutility computing service provider and paybackbones. These are directly connected toonly for the hardware and softwareuser access networks, in order to deliver toresources they use. Utility computing reliesa maximum of users simultaneously, whileheavily on the principle of consolidation,minimizing the path between the user andwhere physical resources are shared by athe desired content.number of applications and users. TheCloud computingprincipal resources offered include, but arenot limitedto,virtual computing Over the years, technology and Internetenvironments (paid per hour and datacompanies such as Google, Amazon,transfer), and storage capacity (paid per Microsoft and others, have acquired aGB or TB used). considerable expertise in operating largedata centers, which are the backbone ofIt is assumed that in-house data centerstheir businesses. Their know-how extendsare idle most of the time due to over-beyond physical infrastructure and includesprovisioning. Over-provisioning is essentialexperience with software, e.g., officeto be sure they can handle peak loadssuites, applications for process(e.g., opening of the trading day or duringmanagement and business intelligence, andholidayshopping seasons),includingbest practices in a range of other domains,unanticipated surges in demand. Utilitysuch as Internet search, maps, email andcomputing allows companies to pay only forother communications applications. In cloudthe computing resources they need, whencomputing, these services are hosted in athey need them.xxviii It also creates marketsdata center and commercialized, so that afor resource owners to sell excesswide range of software applications arecapacities, and therefore make their dataoffered by the provider as a billable servicecenters (and business) more profitable. The(Software as a Service, SaaS) and noexample of online retailer Amazon waslonger need to be installed on the usersmentioned in Box 1. To increase efficiency,PC.xxxiFor example, instead of Outlookone Amazon server can host, in addition tostored on the PC hard drive, Gmail offers aa system managing the companys e-similar service, but the data is stored on Distributed Computing: Utilities, Grids & Clouds (March 2009) 5 8. ITU-T Technology Watch Reports the providers servers and accessed via a approach is to outsource certain tasks to web browser.the cloud, e.g., spam and virus filtering, and to keep other tasks in the corporate For small and medium-sized enterprises, data center, e.g., the storage of the the ability to outsource IT services and mailbox. applications not only offers the potential to reduce overall costs, but also can lower theOther typical services offered include web barriers to entry for many processing-services for search, payment, identification intensive activities, since it eliminates the and mapping. need for up-front capital investment and the necessity of maintaining dedicated Utility and cloud providers infrastructure. Cloud providers gain an The list of providers of utility and cloud additional source of revenue and are able tocomputing services is growing steadily. commercialize their expertise in managing Beside many smaller providers specialized large data centers. cloud and grid services, such as 3teraxxxv, FlexiScalexxxvi, MorphLabsxxxvii, One main assumption in cloud computing xxxviii consists of infinite computing resourcesRightScale , are some of the best known names in web and enterprise computing, of available on demand and delivered via broadband. However that is not always the which three (still) have their core activities in other areas (online retail, Internet case. Problems faced by users in developing countries include the high cost of software search, software) are: xxxix and hardware, a poor power infrastructure,Amazon Web Services (AWS) provide and limited access to broadband. Low-cost companies of all sizes with an computing devicesxxxii equipped with free infrastructure platform in the cloud, and open source software might provide awhich includes computational power, solution for the first problem. Although thestorage, and other infrastructure number of broadband Internet subscribersservices.xl The AWS product range has grown rapidly worldwide, developedincludes EC2 (Elastic Compute Cloud), a economies still dominate subscriptions, and web service that provides computing the gap in terms of penetration incapacity in the cloud, and S3 (Simple developed and developing countries is Storage Service), a scalable storage for wideningxxxiii.Internetusers withoutthe Internet, that can be used to store broadband access are disadvantaged with and retrieve any amount of data, at any respect to broadband users, as they are time, from anywhere on the web. unable to use certain applications, e.g., Google App Engine is a platform for video and audio streaming, online backup of photos and other data. Ubiquitous andbuilding and hosting web applications on unmetered access to broadband Internet is infrastructure operated by Google. The one of the most important requirements forservice is currently in preview, the success of cloud computing. allowing developers to sign up for free and to use up to 500MB of persistent Applications available in the cloud include storage and enough CPU and bandwidth software suites that were traditionally for about 5 million page views a installed on the desktop and can now be month.xli found in the cloud, accessible via a web browser (e.g., for word processing, Salesforce.com is a vendor of communication, email, business intelligence Customer Relationship Management applications, or customer relationship(CRM) solutions, which it delivers using management). This paradigm may save the software as a service model. CRM license fees, costs for maintenance and solutions include applications for sales, software updates, which makes it attractive service and support, and marketing. to small businesses and individuals. Even Force.com is a Platform-as-a-Service some large companies have adopted cloud product of the same vendor that allows solutions with the growing capacities,external developers to create add-on capabilities and success of the service applications that integrate into the CRM providers. Forrester Research suggests that applications and to host them on the cloud-based email solutions would be less vendors infrastructure.xlii expensive than on-premise solutions for upThe Azure Services Platform (Azure) is to 15,000 email accounts.xxxiv Anothera cloud services platform hosted in 9. ITU-T Technology Watch ReportsMicrosoft data centers, which providesliability and QoS can be determined inan operating system and a set ofservice-level agreements.developer services that can be usedSecurity, privacy, anonymity: It may beindividually or together. Afterthe case that the levels of privacy andcompleting its Community Technologyanonymity available to the user of a cloudPreview launched in October 2008, thewill be lower than the user of desktopservices will be priced and licensedapplications.xliv To protect the privacy ofthrough a consumption-based model.xliiicloud users, care must be taken to guardWhile there are different pricing models, the users data and applications forso-called consumption-basedmodels,manipulating that data. Organizations maysometimes referred to as Pay As You Gobe concerned about the security of client(PAYG), are quite popular and measure the dataandproprietary algorithms;resources used to determine charges, e.g.,researchers may be concerned about Computing time, measured in machine unintended releaseofdiscoveries; hoursindividuals may fear the misuse of sensitive Transmissions to and from the datapersonal information. Since the physical center, measured in GB infrastructure in a distributed computing Storage capacity, measured in GBenvironment is shared among its users, any Transactions, measured as application doubts about data security have to be requests overcome.In these types of arrangements, customers Access and usage restrictions: Inare not tied to monthly subscription rates, addition to privacy concerns, the possibilityor other advance payments; they pay onlyof storing and sharing data in clouds raisesfor what they use.concerns about copyright, licenses, andintellectual property. Clouds can beCloud computing and information accessed at any time, by any user with anpolicyInternet connection, from any place.While the main focus of this report is on the Licensing, usageagreementsandimpact of distributed computing on future intellectual property rights may vary instandards work, it should be noted that the different participating countries, but thecontinued and successful deployment ofcloud hides these differences, which cancomputing as a utility presents other cause problems.challenges, including issues of privacy,Governments will need to carefully considersecurity, liability, access, and regulation.the appropriate polices and levels ofDistributed computing paradigms operate regulationorlegislationtoprovideacross borders, and raise jurisdiction andadequatesafeguardsfordistributedlaw enforcement issues similarly to those ofcomputing, e.g. by mandating greaterthe Internet itself. These issues are briefly precisionincontractsandservicedescribed below.agreements between users and providers,Reliability and liability: As with any otherwith a possible view to establishing sometelecommunications service, users willminimal levels of protection. These mayexpect the cloud to be a reliable resource, include:especially if a cloud provider takes over the Basic thresholds for reliability;task ofrunningmission-criticalAssignment of liability for loss or otherapplications, andwill expectclear violation of data;delineation of liability if serious problemsExpectations for data security;occur. Although service disruptions willPrivacy protection;become increasingly rare, they cannot beExpectations for anonymity;excluded.Dataintegrityand the Access and usage rights.correctness of results are other facets ofGartner summarizes seven issues cloudreliability. Erroneous results, data lost orcustomers should address before migratingaltered due to service disruptions can have from in-house infrastructure to externala negative impact on the business of theresources:privileged useraccess,cloud user. The matters of reliability, regulatory compliance, data location, dataDistributed Computing: Utilities, Grids & Clouds (March 2009) 7 10. ITU-T Technology Watch Reports segregation, data recovery, investigativeprogramming interfaces) that improve support, and long-term viability.xlv interoperability by allowing users to adapttheir code and applications according to the While different users (e.g., individuals,requirements of the service. However, the organizations, researchers) may haveAPIs are essentially proprietary and have different expectations for any of thesenot been subject of standardization, which points when they outsource their datameans that users cannot easily extract their and processes to a cloud or grid, it isdata and code from one site to run on necessary that both providers and policyanother. Instead they need to repeat makers address these issues in order toadaptation efforts for each cloud service foster user trust and to handle eventualused. Global standards could allow services events of damage or loss.of different vendors to interoperate. Future standardization workStandardized interfaces would allow usersto use the same code on different Parallels can be drawn between the currentdistributed computing solutions, which state of distributed computing and the earlycould additionally decrease the risk of a days of networking: independent islands oftotal loss of data. systems with little interoperability, only few standards and proprietary management On the provider side, there could be an interfaces:interest in standards for distributed networkmanagement, memory management andThe problem is that theres noload balancing, identity management andstandard to move things around. Isecurity, and standards that allow forthink its the biggest hurdle thatscalability and extensibility of theircloud computing has to face today.infrastructure. Standards work in theseHow do we create an openareas will need to be aware of those whoenvironment between clouds, socontend that such efforts are prematurethat I can have some things resideand could impede innovation.xlviiin my cloud and some things inother peoples data center? A lot ofOther SDOswork needs to be done. Padmasree Among the bodies engaged in theWarrior, CTO, Ciscoxlvi standardization of distributed computing The key to realizing the full benefits ofconcepts are: cloud and grid computing may well lie in Common Componenthttp://www.cca- standardization,particularly in theArchitecture Forumforum.org middleware layer and the area of resource(CCA) interconnection.Distributed http://www.dmtf.org In addition to the questions about Management Task reliability, liability, trust, etc., discussed Force (DMTF) above, the users of distributed computingGlobus Alliance http://www.globus.org infrastructure also are likely to be concernedabout portability and Organization for thehttp://www.oasis- interoperability.Advancement ofopen.orgStructured Portability, the freedom to migrate data onInformation and off the clouds of different providers, Standards (OASIS) without significant effort and switching OpenGridForum http://www.ogf.org costs, should be a major focus of attention(OGF) in standardization.Furthermore, standardized solutions for automation, Optical http://www.oiforum.comInternetworking monitoring, provisioning and configurationForum (OIF) of cloud and grid applications need to be found, in order to provide interoperability. TeleManagementhttp://www.tmforum.org Users may want to employ infrastructureForum (TMF) and services from different providers at the The objective of the Common Component same time. Architecture (CCA) Forum, formed by Todays services include both proprietarymembers of the academichigh- and open source solutions. Many of themperformance computing community, is to providetheir ownAPIs(application definestandardsinterfaces that a 11. ITU-T Technology Watch Reportsframework has to provide to components,Distributed Resource Managementand can expect from them, in order to allowApplication API (DRMAA), a high-leveldisparate components to be interconnected. specification for the submission andSuch standardswouldpromote control of jobs to one or moreinteroperability betweencomponents Distributed Resource Managementdeveloped by different teams acrossSystems (DRMS) within a griddifferent institutions.architecture.The Distributed Management Task ForceConfiguration Description, Deployment,(DMTF) is a group of 160 memberand Lifecycle Management (CDDLM)companies and organizations that developsSpecification, a standard for theand maintains standards for systemsmanagement, deployment andmanagement of IT environments in configuration of grid service lifecycles orenterprises and the Internet. Theseinter-organization resources.standardsenablemanagementThe Globus Alliance is a community ofinteroperability among multi-vendororganizations and individuals developingsystems, tools, and solutions within thefundamental technologies for the grid. Theenterprise in a platform-independent andGlobus Toolkit is an open source gridtechnology-neutral way. DMTF standardsmiddleware component that provides ainclude:standard platform for services to buildCommon Information Model (CIM).upon. The toolkit includes software forDefines how managed elements in an IT security,informationinfrastructure,environment are represented as aresource management, data management,common set of objects and relationships communication, fault detection,andbetween them. This is intended to allow portability.consistent management of theseThe TeleManagement Forum (TMF) is anelements, and to interconnect them,industry association focusedonindependent of their manufacturer ortransforming business processes,provider.operations and systems for managing andWeb-Based Enterprise Managementeconomizingonline information,(WBEM) is a set of standardized systemcommunications andentertainmentmanagement technologies for the services.remote management of heterogeneousExisting Internet standards, such as HTTP,distributed hardware and software XML, SSL/TLS, developed at W3C, IETF,devices.etc. play an important role in thecommunicationbetween client andOpen Virtualization Format (OVF) is anmiddleware.open standard used in the resourcelayer for packaging and distributingITU-Tvirtual appliances or more generallysoftware to be run in virtual machines. The ITU-T has approved a number ofRecommendations that indirectly impact onThe OGF is an open community committeddistributed computing.to driving the rapid evolution and adoptionof applied distributed computing. This is These concern technical aspects, forcritical to developing new, innovative andinstance the work on multimedia coding inscalable applications and infrastructures Study Group 16, or on telecommunicationthat are seen as essential to productivity in security in Study Group 17, as well asthe enterprise and the scientific community.operational aspects, accounting principlesRecommendations developed by the OGFand QoS, treated in Study Groups 2, 3 andcover middleware and resource 12.interconnection layers and includeITU-T Study Groups 13 and 15xlviii haveOpen Grid Services Architectureliaisons with the Optical Internetworking(OGSA), which describes a service-Forum (OIF), which providesoriented grid computing environment interoperability agreements (IAs) thatfor business and scientific use.standardize interfaces for the underlyingDistributed Computing: Utilities, Grids & Clouds (March 2009) 9 12. ITU-T Technology Watch Reports communication infrastructure to enable the APIs. That complicates the process of resources to be dynamically interconnected.moving from one provider to another, or tointernetwork different cloud platforms. ITU-T Recommendations of the E-SeriesIncreased focus on standards for interfaces, (Overall network operation, telephoneand other areas suggested in the report, service, service operation and humanwould enable clouds and grids to be factors) address some of these points andcommoditizedand wouldensure provide, inter alia, definitions related tointeroperability. QoS (E.800) and propose a framework of a Service Level Agreement (E.860). Recommendations in the ITU-T M.3000 series describe the Telecommunication Management Network protocol model, which provides a framework for achieving interconnectivity and communication across heterogeneous operation systems and telecommunication networks. The TMF multi-technologynetworkmanagement solutionisreferencedin ITU-T Recommendation M.3170.0 ff. Conclusion This Report describes different paradigms for distributed computing, namely grid, utility and cloud computing. The spread of communication networks, and in particular the growth of affordable broadband in developedcountries, has enabled organizations to share their computational resources. What originally started as grid computing, temporarilyusingremote supercomputers or clusters of mainframes to address scientific problems too large or too complex to be solved on in-house infrastructures, has evolved into service- oriented business models that offer physical and virtual resources on a pay as you go basis as an alternative to often idle, in- house data centers and stringent license agreements. The user can choose from a huge array of different solutions according to its needs. Each provider offers its own way of accessing the data, often in the form of 13. ITU-T Technology Watch ReportsNotes, sources and further reading Distributed Computing: Utilities, Grids & Clouds (March 2009) 11 14. i http://www.itu.int/ITU-T/studygroups/com17/history.html ii http://gevaperry.typepad.com/main/2007/04/tower_of_babel.html iii Foster, I. and Kesselman, C. The grid: blueprint for a new computing infrastructure, Morgan Kaufmann Publishers Inc., SanFrancisco, CA, 1998 iv http://www.globus.org/alliance/publications/papers/chapter2.pdf v Tanenbaum, A. S. and van Steen, M. Distributed Systems: Principles and Paradigms vi Anderson, R. Security Engineering: A Guide to Building Dependable Distributed Systems, chapter 6 vii Buyya et al. Market-Oriented Cloud Computing: Vision, Hype, and Reality for Delivering IT Services as Computing Utilities,http://www.gridbus.org/papers/hpcc2008_keynote_cloudcomputing.pdf viii http://samj.net/2008/09/taxonomy-6-layer-cloud-computing-stack.html ix http://gridguy.net/?p=10 x Tutschku, K. et al. Trends in network and service operation for the emerging future Internet, Int J Electron Commun (AEU)(2007) xi Delic, K. A. and Walker, M. A. Emergence of the academic computing clouds, Ubiquity 9, 31 (Aug. 2008), 1-1. xii http://www.eu-egee.org/ xiii http://www.cern.ch/ xiv http://www.opensciencegrid.org/ xv http://www.ndgf.org/ xvi http://www.itu.int/ITU-R/conferences/rrc/rrc-06/ xvii http://unosat.org/ xviii ECW is an enhanced compressed wavelet file format designed for geospatial imagery. xix http://www.itu.int/emergencytelecoms xx http://setiathome.berkeley.edu/ xxi http://folding.stanford.edu/ xxii http://lhcathome.cern.ch/ xxiii http://boinc.berkeley.edu/ xxiv http://www.forbes.com/lists/2008/18/biz_2000global08_The-Global-2000_Rank.html xxv http://www.computerweekly.com/Articles/2006/09/26/218593/how-grid-power-pays-off-for-hsbc.htm xxvi http://www.fsg-ship.de/ xxvii http://www.gridipedia.eu/grid-computing-case-studies.html xxviii http://gigaom.com/2008/02/28/how-cloud-utility-computing-are-different/ xxix http://www.akamai.com/ xxx http://www.limelightnetworks.com xxxi Jaeger et al. Cloud Computing and Information Policy: Computing in a Policy Cloud? xxxii infoDev Quick guide: Low-cost computing devices and initiatives for the developing world,http://infodev.org/en/Publication.107.html xxxiii See UNCTAD Information Economy Report 2007-2008, http://www.unctad.org/Templates/webflyer.asp?docid=9479&intItemID=1397&lang=1&mode=highlights, and ITU World Telecommunication/ICT Indicators Database 2008(12th Edition), http://www.itu.int/ITU-D/ict/publications/world/world.html xxxiv http://www.eweek.com/c/a/Messaging-and-Collaboration/SAAS-Email-From-Google-Microsoft-Proves-Cost-Effective-For-Up-to-15K-Seats/1/ xxxv http://www.3tera.com/ xxxvi http://flexiscale.com/ xxxvii http://www.mor.ph/ xxxviii http://www.rightscale.com/ xxxix http://blog.jamesurquhart.com/2008/11/quick-guide-to-big-four-cloud-offerings.html xl http://aws.amazon.com/what-is-aws/ xli http://code.google.com/intl/en/appengine/docs/whatisgoogleappengine.html xlii http://www.salesforce.com/ xliii http://www.microsoft.com/azure/ xliv Delaney Google plans services to store users' data, Wall Street Journal xlv http://www.infoworld.com/article/08/07/02/Gartner_Seven_cloudcomputing_security_risks_1.html xlvi Greenberg (Forbes.com) Bridging the clouds, http://www.forbes.com/technology/2008/06/29/cloud-computing-3tera-tech-cio-cx_ag_0630tera.html xlvii http://samj.net/2008/08/cloud-standards-not-so-fast.html xlviii http://www.oiforum.com/public/liaisons.html