virtualization and cloud computing

Download Virtualization and Cloud Computing

If you can't read please download the document

Upload: laasya

Post on 26-Feb-2016

33 views

Category:

Documents


1 download

DESCRIPTION

Virtualization and Cloud Computing. Data center hardware David Bednárek , Jakub Yaghob , Filip Zavoral. Motivation for data centers. Standardization/consolidation Reduce the number of DCs of an organization Reduce the number of HW, SW platforms - PowerPoint PPT Presentation

TRANSCRIPT

Principy peklada

Virtualization and Cloud ComputingData center hardware

David Bednrek, Jakub Yaghob, Filip ZavoralMotivation for data centersStandardization/consolidationReduce the number of DCs of an organizationReduce the number of HW, SW platformsStandardized computing, networking and management platformsVirtualizationConsolidate multiple DC equipmentLower capital and operational expensesAutomatingAutomating tasks for provisioning, configuration, patching, release management, complianceSecuringPhysical, network, data, user securityData center requirementsBusiness continuityAvailabilityANSI/TIA-942 standardTier 1Single non-redundant distribution pathNon-redundant capacity with availability 99.671% (1729 min/year)Tier 2Redundant capacity with availability 99.741% (1361 min/year)Tier 3Multiple independent distribution pathsAll IT components dual-poweredConcurrently maintainable site infrastructure with availability 99.982% (95 min/year)Tier 4All cooling equipment dual-poweredFault-tolerant site infrastructure with electrical power storage with availability 99.995% (26 min/year)Problems of data centers designMechanical engineering infrastructure designMechanical systems involved in maintaining interior environmentHVAC (heating, ventilation, air conditioning)Humidification and dehumidification, pressurizationSaving space and costs while maintaining availabilityElectrical engineering infrastructure designDistribution, switching, bypass, UPSModular, scalableTechnology infrastructure designCabling for data communication, computer management, keyboard/mouse/videoAvailability expectationsHigher availability needs bring higher capital and operational costsSite selectionAvailability of power grids, networking services, transportation lines, emergency servicesClimatic conditionsProblems of data centers designModularity and flexibilityGrow and change over timeEnvironmental controlTemperature 16-24 C, humidity 40-55%Electrical powerUPS, battery banks, diesel generatorsFully duplicatedPower cablingLow-voltage cable routingCable traysFire protectionActive, passiveSmoke detectors, sprinklers, fire suppression gaseous systemsSecurityPhysical securityProblems of data centers energy useEnergy efficiencyPower usage effectiveness

State of the art DC have PUE 1.2Power and cooling analysisPower is the largest recurring costHot spots, over-cooled areasThermal zone mappingPositioning of DC equipment

Problems of data centers other aspectsNetwork infrastructureRouters and switchesTwo or more upstream service providersFirewalls, VPN gateways, IDSDC infrastructure managementRT monitoring, managementApplicationsDB, file servers, application servers, backupData centers examples

Data centers examples

Data centers examples

Data centers examples

Portable data center

Data centers blade servers

Blade serversModular design optimized to minimize the use of physical space and energyChassisPower, cooling, managementNetworkingMezzanine cardsSwitchesBladeStripped serverStorageStorage area network SANBlock level data storage over dedicated networkServer 1Server 2Switch ASwitch BDiskarray ControlleraControllerbSANServer 1Server 2Switch ASwitch BDiskarray ControlleraControllerbServer nDiskarray ControlleraControllerbDiskarray ControlleraControllerbSAN protocolsiSCSIMapping SCSI over TCP/IPEthernet speeds (1, 10 Gbps)iSERiSCSI Extension over RDMAInfiniBandFCFibre channelHigh speed technology for storage networkingFCoEEncapsulating FC over Ethernet 10High speed4, 8, 16 GbpsThroughput 800, 1600, 3200 MBpsSecurityZoningTopologiesPoint to pointArbitrated loop Switched fabricPortsFCID (like MAC)TypeN node portNL node loop portF fabric portFL fabric loop portE expansion (between two switches)G generic (works as E or F)U universal (any port)NLFibre channelHostStorageNNHostStorageNLNLStorageNLNLNLHostHostSwitchSwitchSwitchStorageStorageNNNNEEFFFFiSCSIInitiatorClientHW, SWTargetStorage resourceLUNLogical unit numberSecurityCHAPVLANLUN maskingNetwork bootingHostInitiator HostInitiator TCP/IP networkDisk arrayTargetABC: A=0, B=1: B=0, C=1FCoEReplaces FC0 and FC1 layers of FCRetaining native FC constructsIntegration with existing FCRequired extensionsEncapsulation of native FC frames into Ethernet framesLossless EthernetMapping FCID and MACConverged network adapterFC HBA+NICConsolidationReduce number of network cardsReduce number of cables and switchesReduce power and cooling costs

FCoE

Disk arraysDisk storage system with multiple disk drivesComponentsDisk array controllersCacheRAM, diskDisk enclosuresPower supplyProvidesAvailability, resiliency, maintainabilityRedundancy, hot swap, RAIDCategoriesNAS, SAN, hybridEnterprise disk arraysAdditional featuresAutomatic failoverSnapshotsDeduplicationReplicationTieringFront end, back endVirtual volumeSpare disksProvisioning

RAID levelsRedundant array of independent disksOriginally redundant array of inexpensive disksWhy?AvailabilityMTBF (Mean Time Between Failure)Nowadays 400 000 hours for consumer disks, 1 400 000 hours for enterprise disksMTTR (Mean Time To Repair)PerformanceOther issuesUsing disks with the same sizeRAID JBODJust Bunch Of DisksMinimum of drives: 1Space efficiency: 1Fault tolerance: 0Array failure rate: 1-(1-r)nRead benefit: 1Write benefit: 1RAID RAID0StripingMinimum of drives: 2Space efficiency: 1Fault tolerance: 0Array failure rate: 1-(1-r)nRead benefit: nWrite benefit: n

RAID RAID1MirroringMinimum of drives: 2Space efficiency: 1/nFault tolerance: n-1Array failure rate: rnRead benefit: nWrite benefit: 1

RAID RAID2Bit striping with dedicated Hamming code parityMinimum of drives: 3Space efficiency: 1-1/n . log2(n-1)Fault tolerance: 1Array failure rate: variableRead benefit: variableWrite benefit: variable

RAID RAID3Byte striping with dedicated parityMinimum of drives: 3Space efficiency: 1-1/nFault tolerance: 1Array failure rate: n(n-1)r2Read benefit: n-1Write benefit: n-1

RAID RAID4Block striping with dedicated parityMinimum of drives: 3Space efficiency: 1-1/nFault tolerance: 1Array failure rate: n(n-1)r2Read benefit: n-1Write benefit: n-1

RAID RAID5Block striping with distributed parityMinimum of drives: 3Space efficiency: 1-1/nFault tolerance: 1Array failure rate: n(n-1)r2Read benefit: n-1Write benefit: n-1

RAID RAID6Block striping with double distributed parityMinimum of drives: 4Space efficiency: 1-2/nFault tolerance: 2Array failure rate: n(n-1)(n-2)r3Read benefit: n-2Write benefit: n-2

RAID nested (hybrid) RAIDRAID 0+1Striped sets in mirrored setMin drives: 4, even number of drivesRAID 1+0 (RAID 10)Mirrored sets in a striped setMin drives: 4, even number of drivesFault tolerance: each mirror can loose a diskRAID 5+0 (RAID50)Block striping with distributed parity in a striped setMin drives: 6Fault tolerance: one disk in each RAID5 blockTieringDifferent tiers with different price, size, performanceTier 0Ultra high performanceDRAM or flash$20-50/GB1M+ IOPS