Transcription
Stretching VMware clustersacross distances with EMC'sVplex - the ultimate in HighAvailability.VMware TechTalk LiveYury Magalif, Principal Architect – Cloud ComputingMarch 11, 2014
What is CDI?Value Added Integrator,Major partners - VMware, EMC, Cisco, HP19 years in business 256 million, 200 employees, 500 clients inNYC Offices in NJ, NYC, Charlotte, Atlanta, Boston, &now Philadelphia #1 EMC Partner NY Metro,Top 5 EMC Nationally #1 VCE Vblock Partner, NY Metro
Customers Want Metro AvailabilityHigh AvailabilityACROSS DATA CENTERSDisaster RecoveryOPERATIONAL AND3RD SITE RECOVERY
How was DisasterAvoidance &Recovery DoneBEFORE StretchedClusters?
Disaster Avoidance & Recovery – HostXVMs move usingVMotion or HA Move VMs using VMotion toAvoid VMs restart using HA to Recover
Disaster Avoidance & Recovery – SiteXVMs move orRecover using SiteRecovery Manager Move VMs using SRM to Avoid VMs restart using SRM to Recover
Traditional Site Recovery ManagervCenterProt.vCenterRecov.vSphere Cluster ADistanceArray-based -typically RecoverPoint,async replication or vSphereReplication (async)Datastore AvSphere Cluster B
How is DisasterAvoidance &Recovery Done WITHStretched Clusters?
VMware High AvailabilityvSphere HA OSAPPOSAPPOSAPPOS VMware High Availability– Extended between distributed parts of the samevirtual datacenter– Automatic rapid recovery from host failures– No complex clustering software in the VM
VMware VM Host AffinityvSphere HA e 1 Affinity Group2APPOSAPPOSAPPOSAPPOSSite 2 Affinity Group VMware Host Affinity– Provides a “site affinity” capability– Keeps workloads local to storage until failure– Keeps primary and secondary FT VMs inappropriate sites
Stretching VMware vSphere ClustersvSphere HA ay basedsynchronousreplicationStretched Storage - EMC VPLEXAPPOSAPPOSAPPOSAPPOSAPPOSAPPOS
Planned Datacenter MigrationvSphere HA OSAPPOSAPPOSAPPOSvMotionMoving all operationsbetween locationsAPPOSAPPOSAPPOSAPPOSAPPOSAPPOSStandard vMotion ofVirtual MachinesStretched Storage - EMC VPLEXAPPOSAPPOSAPPOSAPPOSAPPOSAPPOS
vSphere StretchedClusterConsiderations
Stretched Cluster w/Vplex Requirements 100 km/62 milesThe maximum round trip latency (RTT) 5 ms.Minimum 622 Mbps Vmotion requirementsThe protocol between two VPLEX clusters can beFibre Channel or IP. Layer 2 Stretch (same subnet across both sites) –using classic VLAN or OTV No more than 32 hosts in vSphere cluster (per HA) Protect vCenter with vCenter Heartbeat
Before failing back, disable DRS rules(uncheck BOTH boxes) to control migrationof VMs manually
SRM vs Stretched Cluster Site Recovery Manager with RecoverPoint– Recovery Point Objective 15 minutes to around 1 hour– Recovery Time Objective 1 hour to a couple of hours– Easy button and workflow for recovery– Can build application awareness into workflow– Easy button for DR test -- automatic Stretched Cluster with Vplex–––––Recovery Point Objective 0Recovery Time Objective minutesNo scritable workflow of recoveryHA does not know where application parts are -- manualNo structured DR test – BUT VMotion is easy
What is EMC’s Vplex? Uses same hardware asEMC’s elite array, the VMAX. 3rd site witnessrecommended, but notrequired Typical network bandwidthgenerally 3Gb/s to 10Gb/s Ethernet recommended, butFC supported Capacity based license(recommended)
VMsSite BSite AVMsSite CStretch ClusterIPVirtualCenterStretched LUNSWitnessVNX array w/VplexFC or IPArray atSite-AArray atSite-BNonUniformAccess
vCenterActivevCenter HeartbeatPassive Node
EMC Vplex Customer Examples70% MOREUTILIZATIONALWAYS ON AVAILABILITYACTIVE/ACTIVEDATA CENTERS15%ONLINEMIGRATIONSIMPLEMENTED PRIVATE CLOUDMORE EFFICIENCY83%MIGRATED 250 LIVE SYSTEMSMULTI VENDOR MIGRATIONSLESSMANAGEMENT
Thank you!
Twitter ChatMarch 12, Wednesday2 pm to 3 pm ESTUse Hashtag#CDIVplex
Contact Info & QuestionsFor inquiries about buying CDI solutions, please contactKyle.Mullin@cdillc.com & Adam.Jacobs@CDILLC.comGet the slides in on my work blog:www.cdillc.com/newsroom/cloud-giraffeI would like to thank HP for VC whitepapers/cookbooks from which I borrowed some diagrams in this presentation.For questions after this presentation, email toConnect with me on LinkedIn: Yury ube.com/user/ym640Personal Blog: cloud-zebra.comTechnical Questions:
SRM vs Stretched Cluster Site Recovery Manager with RecoverPoint –Recovery Point Objective 15 minutes to around 1 hour –Recovery Time Objective 1 hour to a couple of hours –Easy button and workflow for recovery –Can build applicat
