Handbook on Data Centers Samee U. Khan • AlbertY. Zomaya Editors Handbook on Data Centers Editors SameeU.Khan AlbertY.Zomaya DepartmentofElectrical SchoolofInformationTechnologies andComputerEngineering TheUniversityofSydney NorthDakotaStateUniversity Sydney Fargo NewSouthWales NorthDakota Australia USA ISBN978-1-4939-2091-4 ISBN978-1-4939-2092-1(eBook) DOI10.1007/978-1-4939-2092-1 LibraryofCongressControlNumber:2014959415 SpringerNewYorkHeidelbergDordrechtLondon © SpringerScience+BusinessMediaNewYork2015 Thisworkissubjecttocopyright.AllrightsarereservedbythePublisher,whetherthewholeorpartofthe materialisconcerned,specificallytherightsoftranslation,reprinting,reuseofillustrations,recitation, broadcasting,reproductiononmicrofilmsorinanyotherphysicalway,andtransmissionorinformation storageandretrieval,electronicadaptation,computersoftware,orbysimilarordissimilarmethodology nowknownorhereafterdeveloped. Theuseofgeneraldescriptivenames,registerednames,trademarks,servicemarks,etc.inthispublication doesnotimply,evenintheabsenceofaspecificstatement,thatsuchnamesareexemptfromtherelevant protectivelawsandregulationsandthereforefreeforgeneraluse. Thepublisher,theauthorsandtheeditorsaresafetoassumethattheadviceandinformationinthisbook arebelievedtobetrueandaccurateatthedateofpublication.Neitherthepublishernortheauthorsorthe editorsgiveawarranty,expressorimplied,withrespecttothematerialcontainedhereinorforanyerrors oromissionsthatmayhavebeenmade. Printedonacid-freepaper SpringerispartofSpringerScience+BusinessMedia(www.springer.com) Preface Data centers play an important role in modern IT infrastructures. A data center is a home to computational power, storage, and applications necessary to support an enterprisebusiness.DatacentersprocessbillionsofInternettransactionseveryday. Nowadays, data centers are the beating hearts of the companies they serve. Large datacenterswiththousandsofservershavebeendeployedbyrenownedICTorgani- zations,likeIBM,Microsoft,Amazon,Yahoo,E-Bay,andGoogle,toprovidecloud computingservices.Thephenomenalincreaseinthesizeandnumberofdatacen- tersandresultantincreaseinoperationalcosthasstimulatedtheresearchinvarious domainsofdatacenters,suchasenergyefficiency,resourcemanagement,network- ing,andsecurity.ThisdetailedHandbookonDataCenterscoversinasuccinctand orderlymannerallaspectspertainingtodatacenterstechnologies. Giventhefastgrowingexpansionofdatacenters,itisnotsurprisingthatavariety ofmethodsarenowavailabletoresearchersandpractitionersstrivingtoimprovedata center performance. This handbook aims to organize all major concepts, theories, methodologies, trends, challenges and applications of data centers into a coherent andunifiedrepository.Moreover,thehandbookprovidesresearchers,scholars,stu- dents and professionals with a comprehensive, yet concise source of reference to datacentersdesign,energyefficiency,resourcemanagement,andscalabilityissues. The handbook consists of nine parts, where each part consists of several chapters. Thefollowingtopicsspanningthedatacentertechnologiesarecoveredindetail:(a) energyefficiency,(b)networking,(c)cloudcomputing,(d)hardware,(e)modeling andsimulation,(f)security,(g)dataservices,(h)monitoring,and(i)resourceman- agement.Eachpartdescribesthestateoftheartmethods, aswellastheextensions andnovelmethodologiesdevelopedrecentlyfordatacenters. Thefirstpartdescribesenergyefficiencymethodsfordatacentersandaddresses thedesigningofenergyefficientschedulingmechanismsforhighperformancecom- putingenvironments.Theauthorshavediscussedenergy-awarealgorithmsfortask graphscheduling, replicaplacement, check-pointing, numericalprocessing, andto estimateenergyconsumptionforthegivenworkload.Also, thispartaddressesthe problemofminimizingenergyconsumptionwithschedulelengthconstraintonmul- ticore processors. The following chapters covers the energy efficiency aspects of datacenterspresentedinthehandbook:(a)Energy-EfficientandHigh-Performance v vi Preface ProcessingofLarge-ScaleParallelApplicationsinDataCenters,(b)Energy-aware algorithmsfortaskgraphscheduling,replicaplacementandcheckpointstrategies, (c) Energy efficiency in HPC Data Centers: Latest Advances to Build the Path to Exascale,(d)Techniquestoachieveenergyproportionalityindatacenters:asurvey, (e) A Power-Aware Autonomic Approach for Performance Management of Scien- tificApplicationsinaDataCenterEnvironment,(f)CoolEmAll:ModelsandTools forPlanningandOperatingEnergyEfficientDataCentres, (g)SmartDatacenter, (h) Power and Thermal Efficient Numerical Processing, and (k) Providing Green ServicesinHPCDataCenters:AMethodologybasedonEnergyEstimation. Thesecondpartofthehandbookprovidesastudyofvariouscommunicationand networkingmethodologiesfordatacenters.Thenetworkvirtualizationconcepts,the architectureofopticalnetworksfordatacenters,andnetworkscalabilityissuesare discussed in detail. Moreover, an emphasis is drawn over packet classification in multicoreplatforms.Theroutingtechniquesfordatacenternetworksarediscussed indetailalongwithstudyonTCPcongestioncontrolindatacenternetworks.The followingchaptersprovideadetailedoverviewintermsofnetworkingtechnologies fordatacentersareincludedinthispart:(a)NetworkVirtualizationinDataCenters: A Data Plane Perspective, (b) Optical data center networks: Architecture, perfor- mance,andenergyefficiency,(c)ScalableNetworkCommunicationusingUnreliable RDMA,(d)PacketClassificationonMulti-corePlatforms,(e)OpticalInterconnects for Data Center Networks, (f)TCP Congestion Control in Data Center Networks, and(g)RoutingTechniquesinDataCenterNetworks. Thethirdpartofthehandbookdiscussestheroleofdatacentersincloudcomput- ingandhighlightsvariouschallengesfacedinensuringthedataintegrity,reliability, and privacy in cloud computing environments. The role of trusted third parties in performingmonitoringandauditingofservicelevelagreementsisillustrated,aswell as integrity of big data in cloud computing is discussed. The data intensive appli- cationsincloudarediscussedalongwithstoragechallenges.Thispartincludesthe followingchapters:(a)AuditingforDataIntegrityandReliabilityinCloudStorage, (b)I/OandFileSystemsforData-IntensiveApplications, (c)Cloudresourcepric- ingundertenantrationality,(d)OnlineResourceManagementforCarbon-Neutral Cloud Computing, (e)A Big Picture of IntegrityVerification of Big Data in Cloud Computing,(f)AnOut-of-CoreTask-basedMiddlewareforData-IntensiveScientific Computing,(g)BuildingScalableSoftwareforDataCenters:AnApproachtoDis- tributedComputingatEnterpriseLevel,and(h)CloudStorageoverMultipleData Centers. Thefourthpartofthehandbookshedalightondatacentersemerginghardware technologies.Theissuespertainingtotheefficientdatastorageonredundantarray ofindependentdisks(RAID)technologies, thedatasynchronizationchallengeson many-cores, and hardware approaches to transactional memory on chip multipro- cessorsaredetailed.Thefollowingchaptersareincludedinthispart:(a)Realizing Accelerated Cost-Effective Distributed RAID, (b) Efficient Hardware-Supported Synchronization Mechanisms for Many-cores, and (c) Hardware Approaches to TransactionalMemoryinChipMultiprocessors Preface vii The fifth part of the handbook discusses modeling and simulation techniques for data centers and include following chapters: (a) Data Center Modeling and SimulationUsingOMNeT++,(b)Power-ThermalModelingandControlofEnergy- EfficientServersandDatacenters,(c)Thermalmodelingandmanagementofstorage systemsindatacenters,and(d)ModelingandSimulationofDataCenterNetworks. This portion highlights various techniques to model power consumption, thermal response,storagesystems,andcommunicationindatacenternetworks.Theauthors utilizeddiscrete-timesimulatorOMNet++tomodelthedatacentertraffic. The sixth part of the handbook provides a discussion on various security and privacytechniquesfordatacenters.Amodelisproposedtodetectandmitigatethe covertchannelsindatacenters.Theprivacyissuesregardingdatacenteroutsourcing are discussed and a survey of various privacy attacks and their counter measures is presented. The following chapters are included in this portion: (a) C2Hunter: Detection and Mitigation of Covert Channels in Data Centers, (b) Selective and Private Access to Outsourced Data Centers, and (c) Privacy in Data Centers: A SurveyofAttacksandCountermeasures. The seventh part details the data services and their management in data center. Thequalityofservicerequirementsforprocessingstreamdataforcityapplicationsis discussed.Thedatamanagementandqueryingofbigdataissurveyedindetail.The authors also discussed the various methods of constructing on the fly data centers in wireless ad hoc network environments. The following chapters are included in this part: (a) Quality-of-Service in Data Center Stream Processing for Smart City Applications, (b) Opportunistic Databank: A context aware on-the-fly data center for mobile networks, (c) Data Management:State-of-the-Practice at Open-Science DataCenters,and(d)DataSummarizationTechniquesforBigData—ASurvey. Theeighthpartofhandbookillustratesvarioushardwareandsoftware-basedmon- itoringsolutionsfordatacenters.Theuseofwirelesssensornetworkstechnologyto sensethethermalactivitiesandefficientcirculationofcoolingairisexplored.Traffic monitoringisanothervitaltopicdiscussedwithasurveyonnetworkintrusiondetec- tionsystemsfordatacenters.Thefollowingchaptersarecoveredinthisportion:(a) CentralManagementofDatacenters,(b)MonitoringofDataCentersusingWireless SensorNetworks,(c)NetworkIntrusionDetectionSystemsinDataCenters,and(d) SoftwareMonitoringinDataCenters. Lastly,theninthpartofbookchapterelaboratestheresourcemanagementaspects of data centers and present a detailed case study on usage patterns in multi-tenant datacenters.Adiscussionisalsoperformedonschedulingofdistributedtransactional memoryandthevarioustradeoffsandtechniquesindistributedmemorymanagement areelaboratedindetail.Moreover,anemphasisisdrawnontheresourcescheduling in data-centric systems. This part includes the following chapters: (a) Usage Pat- terns in Multi-tenant Data Centers: a Large-Case Field Study, (b) On Scheduling inDistributedTransactionalMemory:TechniquesandTradeoffs,(c)Dependability- OrientedResourceManagementSchemesforCloudComputingDataCenters,and (d)ResourceSchedulinginData-CentricSystems. Contents PartI EnergyEfficiency Energy-Efficient and High-Performance Processing of Large-Scale ParallelApplicationsinDataCenters................................ 3 KeqinLi Energy-Aware Algorithms for Task Graph Scheduling, Replica PlacementandCheckpointStrategies................................ 37 GuillaumeAupy,AnneBenoit,PaulRenaud-GoudandYvesRobert Energy Efficiency in HPC Data Centers: LatestAdvances to Build thePathtoExascale ............................................... 81 SébastienVarrette,PascalBouvry,MateuszJarusandArielOleksiak Techniques toAchieve Energy Proportionality in Data Centers: ASurvey ......................................................... 109 MadhurimaPore,ZahraAbbasi,SandeepK.S.Gupta andGeorgiosVarsamopoulos A Power-AwareAutonomicApproach for Performance Management ofScientificApplicationsinaDataCenterEnvironment ............... 163 RajatMehrotra,IoanaBanicescu,SrishtiSrivastava andSherifAbdelwahed CoolEmAll: Models andTools for Planning and Operating Energy EfficientDataCentres ............................................. 191 MichavordemBerge,JochenBuchholz,LeandroCupertino, GeorgesDaCosta,AndrewDonoghue,GeorginaGallizo,MateuszJarus, LaraLopez,ArielOleksiak,EnricPages,WojciechPiatek, Jean-MarcPierson,TomaszPiontek,DanielRathgeb,JaumeSalom, LauraSisó,EugenVolk,UweWössnerandThomasZilio ix x Contents SmartDataCenter ................................................ 247 MuhammadUsmanShahidKhanandSameeU.Khan PowerandThermalEfficientNumericalProcessing ................... 263 WeiLiuandAlbertoNannarelli ProvidingGreenServicesinHPCDataCenters:AMethodologyBased onEnergyEstimation.............................................. 287 MohammedElMehdiDiouri,OlivierGlück,LaurentLefêvre andJean-ChristopheMignot PartII Networking NetworkVirtualizationinDataCenters:ADataPlanePerspective ...... 327 WeirongJiangandViktorK.Prasanna OpticalDataCenterNetworks:Architecture,Performance,andEnergy Efficiency ........................................................ 351 YuichiOhsitaandMasayukiMurata ScalableNetworkCommunicationUsingUnreliableRDMA............ 393 RyanE.Grant,MohammadJ.Rashti,PavanBalajiandAhmadAfsahi PacketClassificationonMulti-corePlatforms ........................ 425 YunR.Qu,ShijieZhouandViktorK.Prasanna OpticalInterconnectsforDataCenterNetworks ...................... 449 KhurramAzizandMohsinFayyaz TCPCongestionControlinDataCenterNetworks .................... 485 RohitP.Tahiliani,MohitP.TahilianiandK.ChandraSekaran RoutingTechniquesinDataCenterNetworks ........................ 507 ShaistaHabib,FawazS.BokhariandSameeU.Khan PartIII CloudComputing AuditingforDataIntegrityandReliabilityinCloudStorage ........... 535 BingweiLiuandYuChen I/OandFileSystemsforData-IntensiveApplications .................. 561 YanlongYin,HuiJinandXian-HeSun CloudResourcePricingUnderTenantRationality .................... 583 XinJinandYu-KwongKwok Contents xi OnlineResourceManagementforCarbon-NeutralCloudComputing ... 607 KishwarAhmed,ShaoleiRen,YuxiongHeandAthanasiosV.Vasilakos A Big Picture of Integrity Verification of Big Data in Cloud Computing ....................................................... 631 ChangLiu,RajivRanjan,XuyunZhang,ChiYangandJinjunChen AnOut-of-CoreTask-basedMiddlewareforData-IntensiveScientific Computing ....................................................... 647 ErikSaule,HasanMetinAktulga,ChaoYang,EsmondG.Ng andÜmitV.Çatalyürek Building Scalable Software for Data Centers: An Approach toDistributedComputingatEnterpriseLevel ........................ 669 FernandoTurradoGarcía,AnaLucilaSandovalOrozco andLuisJavierGarcíaVillalba CloudStorageoverMultipleDataCenters ........................... 691 ShuaiMu,MaomengSu,PinGao,Yongwei Wu,KeqinLi andAlbertY.Zomaya PartIV Hardware RealizingAcceleratedCost-EffectiveDistributedRAID................ 729 AleksandrKhasymski,M.MustafaRafique,AliR.Butt, SudharshanS.VazhkudaiandDimitriosS.Nikolopoulos Efficient Hardware-Supported Synchronization Mechanisms forManycores .................................................... 753 JoséL.Abellán,JuanFernándezandManuelE.Acacio Hardware Approaches to Transactional Memory in Chip Multiprocessors................................................... 805 J.RubénTitos-GilandManuelE.Acacio PartV ModelingandSimulation DataCenterModelingandSimulationUsingOMNeT++.............. 839 AsadW.MalikandSameeU.Khan Power-Thermal Modeling and Control of Energy-Efficient Servers andDatacenters................................................... 857 JungsooKim,MohamedM.Sabry,MartinoRuggieroandDavidAtienza