Above the Clouds A Berkeley View of Cloud ComputingOutlineWhat is Cloud Computing?Why Now?Spectrum of CloudsCloud Killer AppsEconomics of Cloud UsersSlide 8Slide 9Economics of Cloud ProvidersAdoption ChallengesGrowth ChallengesPolicy and Business ChallengesShort Term ImplicationsLong Term ImplicationsUC BerkeleyAbove the CloudsA Berkeley View of Cloud Computing1UC Berkeley RAD LabOutline•What is it?•Why now?•Cloud killer apps•Economics for users•Economics for providers•Challenges and opportunities•Implications2What is Cloud Computing?•Old idea: Software as a Service (SaaS)–Def: delivering applications over the Internet•Recently: “[Hardware, Infrastrucuture, Platform] as a service”–Poorly defined so we avoid all “X as a service”•Utility Computing: pay-as-you-go computing–Illusion of infinite resources–No up-front cost–Fine-grained billing (e.g. hourly) 3Why Now?•Experience with very large datacenters–Unprecedented economies of scale•Other factors–Pervasive broadband Internet–Fast x86 virtualization–Pay-as-you-go billing model–Standard software stack4Spectrum of Clouds•Instruction Set VM (Amazon EC2, 3Tera)•Bytecode VM (Microsoft Azure)•Framework VM–Google AppEngine, Force.comEC2 Azure AppEngine Force.comLower-level,Less managementHigher-level,More management5Cloud Killer Apps•Mobile and web applications•Extensions of desktop software–Matlab, Mathematica•Batch processing / MapReduce–Oracle at Harvard, Hadoop at NY Times6Unused resourcesEconomics of Cloud Users•Pay by use instead of provisioning for peakStatic data center Data center in the cloudDemandCapacityTimeDemandCapacityTime7Unused resourcesEconomics of Cloud Users•Risk of over-provisioning: underutilizationStatic data centerDemandCapacityTime8Economics of Cloud Users•Heavy penalty for under-provisioningLost revenueLost usersDemandCapacityTime (days)12 3DemandCapacityTime (days)12 3DemandCapacityTime (days)12 39Economics of Cloud Providers•5-7x economies of scale [Hamilton 2008]•Extra benefits–Amazon: utilize off-peak capacity–Microsoft: sell .NET tools–Google: reuse existing infrastructureResourceCost inMedium DCCost inVery Large DCRatioNetwork $95 / Mbps / month $13 / Mbps / month 7.1xStorage $2.20 / GB / month $0.40 / GB / month 5.7xAdministration ≈140 servers/admin >1000 servers/admin 7.1x10Adoption ChallengesChallenge OpportunityAvailability Multiple providers & DCsData lock-in StandardizationData Confidentiality and AuditabilityEncryption, VLANs, Firewalls; Geographical Data Storage11Growth ChallengesChallenge OpportunityData transfer bottlenecksFedEx-ing disks, Data Backup/ArchivalPerformance unpredictabilityImproved VM support, flash memory, scheduling VMsScalable storage Invent scalable storeBugs in large distributed systemsInvent Debugger that relies on Distributed VMsScaling quickly Invent Auto-Scaler that relies on ML; Snapshots12Policy and Business ChallengesChallenge OpportunityReputation Fate Sharing Offer reputation-guarding services like those for emailSoftware Licensing Pay-for-use licenses; Bulk use sales13Short Term Implications•Startups and prototyping•One-off tasks–Washington post, NY Times•Cost associativity for scientific applications•Research at scale14Long Term Implications•Application software:–Cloud & client parts, disconnection tolerance•Infrastructure software:–Resource accounting, VM awareness•Hardware systems:–Containers, energy
View Full Document
Unlocking...