J 61 FEATURE G RID C OMPUTING ID: 506249
Download Pdf The PPT/PDF document "Can computational grids make as great an..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
J 61 FEATURE G RID C OMPUTING ! !"! # $ # $ ! % "& ! "& ! ' ! ' !" ! ! ' ! *! ! ! ! ! ! " ! !+ ! ! ' %, ' -! !. ! ! ! ' ! ", ! ) ! 1$ Can computational grids make as great an impact in the 21st century as electrical gridsdid in the 20th? A comparison of the two technologies could provide clues about how tomake computational grids pervasive, dependable, and convenient. 6 7== 1521-9615/02/$17.00 © 2002 IEEE 62 OMPUTINGIN ! Computational grids , ! ! + A 4 , !B C " ! " ) ! % ! ! ! 9 ! ) ! % ' E - '! ! ! 'E . % F ) ! ) ! *! ! $ & ! " Technological milestones , ! ! + ! ! !! $ + Figure 1. Volta demonstrates the battery for Napoleon I at theFrench National Institute, Paris, in 1801. The painting is from theNational History Museum, Florence University, Italy. Applications of Grid Computing operations research,processed a database of large pulsar signals in a search for and aggregating resources to solve problems. References 1.D. Abramson, J. Giddy, and L. Kotler, IntÕl Parallel and Distributed Processing Symp. 2.R. Buyya et al., J. Con-currency and Computation: Practice and Experience 3.A. Oram, ed., Peer-to-Peer: Harnessing the Power of Disruptive Technolo-gies 4.I. Foster, C. Kesselman, and S. Tuecke, Enabling Scalable Virtual Organizations, Intl J. Supercomputer Applica-tions 5.R. Buyya et al., Proc. SPIE IntÕl Conf. on Commercial Appli- , SPIE, Bellingham, Wash.,2001. power grid..... J 63 !"! /! C ' ! " Communication. ! ! !! $ ' ! 7;: : 9 %C ": 9% 7;;9 !; ! ' (GH" ; : (I(: 9% ! ! + 7 , 8 2; 9 $ ! 27 , ; 3 2 7 7 "3 (H $ ' 0! ! ) ! ! !"$(H036 ! ' ! ) (HI"$(HD, 3@ ,%J$%(H"$(0 ! ! ' "$(G # ; ! : 9% ! ! ! ! "$(I $ ' -$ . E ! ! !$ "$(( #;"6! $ *! ! ! " & + (( 3 8 , 9#;E +! C !$ ! E ! 2 ' ! " & & & , ! -"0". ((D " ) 'K68- ) 6 '! 8 ! . ' ! $ " Computation. !! ,%7 (H ! ' ' "-; : ! , ! A: ! 4 , !& ; B ". ) ! ! (H0 K )% 9 , '"; ¥ Ethernet¥ IETF¥ Internet era¥ WWW era¥ Mosaic¥ XML¥ PC clusters¥Crays¥MPPs¥ HTML¥ W3C¥Xerox Parc worm¥ Web services ¥ Minicomputers¥ ¥ WS clusters¥ PDAs¥ Workstations¥ HTC Networking advances have come the rise and fall of various systems. In the 1960s, mainframes (mainly from IBM) served the needs ofcomputing users, but a decade later DECthe 1980s, vector computers such as Crays and, later, parallel computers such as massively parallel processors became the 64 OMPUTINGIN % 9, ! ! " ! ! ! ! ! '"& ! ! ! "; (( ! ! " !%, ' ' - ! . ! !" !! E ! ! ! ! ", ! ! ! ! ! ! ! %/% ! ! %, $ "%/% ' ! ; $L C # 4! "4 ! %/% C ! / ! " Layered structure ! + ! ! "! 0 ! ! "-$ , @ ; ! ' ! ! " ' ! ! ! ! ' " ! " , ! ! %, ' ;6% - ! .! ! 7)& ,! ! ! 9 ! ! 8 ; , % 3 ; ;!4 ; : ; ! ! ) ! !E ! " ! Distributed and Grid Computing Web Sites Distributed.net, Project RC5 www.distributed.net/rc5 Global Grid Forum www.gridforum.org Grid Computing Info Centre www.gridcomputing.com IEEE Distributed Systems Online http://dsonline.computer.org History of Distributed Computing www.ud.com/company/dc/history.htm Internet Timeline www.zakon.org/robert/internet/timeline Peer-to-Peer Working Group www.p2pwg.org SETI@home http://setiathome.ssl.berkeley.edu Grid applicationsScience, engineering, commercial applications, Web portalsGrid programming environments and toolsLanguages, interfaces, libraries, compilers, parallelization toolsCore grid middlewareJob submission, storage access, info services, trading accountingGrid fabricPCs, workstations, clusters, networks, software, databases, devicesSecurity infrastructureSingle sign-on, authentication, secure communcationUser-level middlewareÑresource aggregatorsResource management and scheduling services Figure 3. A layered architecture for the computational grid and J 65 ! ! - ! ." ) ! ' ! ! 2 9 ! ' ) ! ! ! ! ! ": ! " ! M ! ! " Operational model ! ' ! ! ! ! ! ' - . ! " ' ) ! !! ! ! - ! D."; C ) ! ! " ! ! - 5; *! ! . M *! - ,%7 $J5 ' ." - ! . - ! . C "6 ! ! ! ! E "9 ! ! 2 E E " ! E ! " E ! " ! ) ! ! ! M ) ! " E ! ! "& ) ! ! ! " ! ' ! ! "% ! ! ! "5 ! ! ! !E ! ! !E ! "5 ! ! ! ! ) ! " ! ' ) ! ! " ! ' ' ! Grid information serviceGrid information serviceApplicationRNR1R5R6Grid resource broker Database Grid resource broker R3 2 R4 view of a 66 OMPUTINGIN 2 ) ! ! " Comparing the grids ! ' ! ! ! " *! ! M "$ ! C !C Table 1. Electrical and computational power grids: A comparison. ParameterElectrical power gridComputational power grid ResourcesHeterogeneous: thermal, hydro,Heterogeneous: PCs, workstations, clusters, and others; wind, solar, nuclear, othersdriven by different operating and management systems NetworkTransmission lines, undergroundInternet is the carrier for connecting distributed cables. Various sophisticated resorces, load, and so on. Analogous quantitiesBusNodeEnergy transmissionComputational transmissionVoltageBandwidthBulk transmission system Bulk transmission by beroptic-OC48, ATM (2.4 Gbps)Subtransmission (25 kV to 150 kV)Ethernet, T-3 (45 Mbps)Distribution (120/240V, 25 kV)Modem, ISDN, and so on (56 to 128 Kbps)CableCableEnergy (MW-hour)Computational power (MOnly small storage capacity in theAny magnitude of storage (Mbytes) Power sourcePower station (turbogenerators,Grid resource (computers, data sources, Web services,hydrogenerators), windmilldatabases) Load type Heterogeneous application devices: Heterogeneous applications: for example, graphics for(based on use type)for example, mechanical energy for multimedia applications, problem solving for scientifans, electricity for TVs, heat for irons or engineering applications Operating frequencyUniform: 50 or 60 HzNonuniform: Depends on computer processing powerDC systems also existand clock speed.Analog quantity, sinusoidalDigital, square wave Access interfaceDirect: Wall socket for small consumers,Uniform interface to heterogeneous resources: for examGlobus GRAM interface for submitting jobs to resources Ease of useVery simple: Plug and playVery complex: Expected to change as computing portals Matching device toTransformer changes voltage levels toResource brokers select resources to meet user varying power levelsmatch, for example, a 25 V devicerequirements such as quality and cost. Applications can (voltage, bandwidth,with a 220 V supply.run on machines with different capabilities, so devicesCPU speed)like transformers aren Aggregation of When a load requires more power When an application needs more computational powerresourcesthan can be provided locally, the gridthan a single resource can provide, or for faster provides additional power. Economic execution, computational grids allow resourcedispatch center uses sophisticated aggregation for executing application components inow parallel. Grid resource brokers such as Nimrod-Gstudies that provide the mechanismsprovide resource aggregation capability.to carry this out. ReliabilityImportant lines are duplicated. Resources in a grid may fail without notice. ResourceSophisticated protection schemes exist brokers must handle such failure issues at runtime. J 67 *! " ) ! + ) ! ! ' "$ ! ";! ! ! ! "3 !! ! " ParameterElectrical power gridComputational power grid StabilityStability is crucial for keeping the Stability depends on resource management policy.generators in sync. Sophisticated If resource is shared, available computing power for acontrol algorithms ensure automated user can vary. Transmission capacityMaximum upper limit for the lines Upper limit depends on carriers bandwidth capability. thermal limits. Security/safetyFuses, circuit breakers, and so onFirewalls, public-key infrastructure, and PKI-based grid Cogeneration OptionalOptional StorageOnly storage for low-power DC No storage of computational power is possible. Automated accountingAdvanced metering and accountingLocal resource management systems support mechanisms are in place.accounting. Resource brokers can meter resource metering); global-level service exchange and accounting InterconnectionVarious regional power pools are Internet provides connectivity service; tools such asinterconnected by weak connectionsJobQueue in Legioncalled tie-lines. federation resources with tight coupling. Unregulated grid Successful operation in countries withNot yet. As this technology matures and businesses startoperationsufcient generation capacity. Regulated grid Load dispatch center manages optimalGreater potential exists for using market-based pricingoperationsystem operation.mechanisms to help regulate resource supply and RegulatorsIn general, managed by an auto- No regulator yet exists. However, the need for anomous body of vendors and watchdog will grow as the grid enters mainstreamfor example,computing. Some national supercomputing centersNEMMCO in Australia (www.(for example, in the UKnemmco.com.au).committee that decides on token allocation resource. This resembles price regulation in a single Standards bodyMany standardization bodies existForums such as Global Grid Forum and the P2P Workingfor various components, devices, Group promote community practices. The IETF andsystem operation, and so on. (For W3C handle Internet and Web standardization issues. 68 OMPUTINGIN ! ) ) ! " Resources 2G 2 G ! " ! ! + + ! "6 M ! + ! ! ! !1 !! + ! - ! . GI; ! ! ! ";! ! ! ;6% ! ! %, ' ! ! "$ ! E M ! - ! . ! ' ! ! - ! D.", ! " " ! ! ' ! " Network ! ' )" "3 ' ! ' " ! G ' E 1 ! !" ' C " ! ! ' C " ! ! ' " ! ! ! ! M - ! ! . '2 ! ! "$ ! ! - . $ ! ! 8 # ! ' ! M ! " 8 # 8 # & #" ' - ! . ! ' - ! ." ! !' ! ' 6 "; ! ! ! 2 ! " System load ! ! ' ! "; ! ! - ! ) . ! - $J5 ! ) ." ! ' ) ! " *! ! ! ! " !' ! ! ! ! ! ! " Operational model & ! ! J 69 !*!! " "$ ! *! " ( ! !! ! ! M "& ! ' !! ! " *! ! ! ! "5! ! ! ! "3! ! ! ! " *! "$ ) ! ! "6 F ! *! ! " ! ! ! *! ! "; ( !! ! " ! + ! + " ' ' ' " ! ! ! ! + ) ! ! E - . ! ! ! ! ! " ! ' "$ ! ! ' ! ! ! "3 ' " M ) ' " ! *! ! !EE " *! "! ! - ) . ! ! ! " Dissimilarities in the two grids 5! ! ", ! " ) ! -C! . Primary distribution G: Sync. Generator Subtransmissionlevel Transmission level Secondary distribution Primary distribution Subtransmissionlevel Secondary distribution Very large consumer To other pool member : Transformer Figure 5. Power system diagram. 70 OMPUTINGIN ' "% C ! " ! ! "! , ! :,- ! ! " ! ' *! ! " ! ! ! ! I"% ! ! K M ! " ! ! ! ! ! ! ! ";! ) ! !! ! "; ) ! , 4 8 ?N! ! ! "6 ! ! ! ! *! ! ! " ! ! - !. " ' ! ' ! ! ! ! E ' ? ! ! " ) 2 ! ! " ! ' ! " ! ! E M !! "%! ! *! C ! "5! ! ! " . ! ! ' ! ! ! Acknowledgments We thank Domenico Laforenza, Ajith Abraham, RobGray, David Walker, Alexander Reinefeld, and FrankKarbarz for their constructive comments. We thank DavidAbramson for his encouragement and support. Weextend special thanks to Domenico Laforenza forproviding us a copy of the photograph included in Figure1. Computational grid content is derived from Buyya References 1.I. Foster and C. Kesselman, eds., The Grid: Blueprint for a FutureComputing Infrastructure 2.A. Raman et al., Proc. Third IntÕl Conf. High-Perfor- gridgrid Centralgrid grid grid Figure 6. A schematic overview of the three levels of grid. J 71 mance Computing 3.R. Buyya, ed., High Performance Cluster Computing: Architectures 4.R. Buyya, PhD thesis, Monash Univ.,5.A. Oram, ed., Peer-to-Peer: Harnessing the Power of DisruptiveTechnologies 6.R. Metcalfe and D. Boggs, Ethernet: Distributed Packet Switch- Proc. ACM Nat 7.S. Harris, 2001; www.ietf.cnri.reston.va.us/rfc/8.T. Berners-Lee, the World Wide WebBy Its Inventor 9.R. Buyya, ed., High Performance Cluster Computing, 10.M. Baker, R. Buyya, and D. Laforenza, Proc. Intl Conf. Advances in Infra-structure for Electronic Business, Science, and Education on the In- 11.I. Foster, C. Kesselman, and S. Tuecke, Intl J. SupercomputerApplications 12.R. Buyya, D. Abramson, and J. Giddy, Proc. Fourth IntÕl Conf. High-Perfor-mance Computing 13.R. Buyya, D. Abramson, and J. Giddy, Proc.l Conf. Parallel and Distributed Processing Techniques and Ap-plications 14.R. Buyya et al., Economic Models for Management of Resources Proc. SPIE IntÕl Conf. Com-mercial Applications for High-Performance Computing 15.O. Elgerd, Electric Energy Systems Theory: An Introduction 16.I. Foster and C. Kesselman, l J. Supercomputer Applications 17.H. Casanova and J. Dongarra, Intl J. SupercomputerApplications and High Performance Computing 18.I. Foster et al., Proc. 5th ACM Conf. Computer and Communications Security 19.R. Buyya, J. Giddy, and D. Abramson, l Heterogeneous Computing Workshop 20.D. Katramatos et al., JobQueue: A Computational Grid-Wide Proc. 2nd IntÕl Workshop Grid Computing 21.J. Frey et al., Proc. 10th Intl Symp. High-Perfor- 22.J. Brooke et al., Mini-Grids: Effective Test-beds for Grid Appli- Proc. 1st IEEE/ACM IntÕl Workshop Grid Computing 23.A. David and F. Wen, Proc. IEEE Power Engineering So-ciety Summer Meeting 24.J. Lamont and S. Rajan, IEEE Trans. Power Systems 25.R. Ferrero, S. Shahidehpur, and V. Ramesh, IEEETrans. Power Systems 26.Z. Younes and M. Ilic, Decision Support System 27.P. Myrseth, Proc. OECD Workshop on Business-to-Business Elec-tronic Commerce: Status, Economic Impact and Policy Implications www.nr.no/~pmyrseth/artikler/oecd_ie_wokshop_99; currentMarch 2002. Madhu Chetty Computing and Information Technology, Monash Uni-versity, Churchill, Australia. His current research inter-ests are resource scheduling in computational grids,cialin-neering from Nagpur University, India, and has morethan 20 years of teaching and research experience. Con- Rajkumar Buyya is leading the Grid Computing andDistributed Systems Laboratory in the School of Com-Melbourne, Australia. His research interests includesoftware for the PARAM supercomputers produced by gramming, Mastering C++, and Design of PARAS Micro- , and the editor of High Performance Cluster Dharma Ratnakara Memorial Trust Gold Medal for aca-topic, please visit our digital library at http://computer.