Event Title
High Availability withHigh Availability with
Windows FailoverWindows Failover
Clustering and Geo-Clustering and Geo-
ClusteringClustering
Michael Otey
Senior Technical Director
Windows IT Pro
SQL Server Pro
1
Event Title
What this presentation coversWhat this presentation covers
 Today’s IT requirementsToday’s IT requirements
 The costs of downtimeThe costs of downtime
 Understanding Windows FailoverUnderstanding Windows Failover
ClusteringClustering
 New features in Ws2012 / R2 clusteringNew features in Ws2012 / R2 clustering
 Geo-clustersGeo-clusters
 Architecting availabilityArchitecting availability
 Additional referencesAdditional references
2
Event Title
Today’s IT RequirementsToday’s IT Requirements
 AgilityAgility
 Meeting quickly changing business conditionsMeeting quickly changing business conditions
 Reducing complexityReducing complexity
 Increasing automationIncreasing automation
 Consumerization of ITConsumerization of IT
 Reducing costsReducing costs
 CapEXCapEX
 OpExOpEx
 Data control and securityData control and security
Event Title
The Costs of DowntimeThe Costs of Downtime
 GartnerGartner
 $5600 p/m$5600 p/m
 $300K p/h$300K p/h
 Amazon 30 minute outage in 2013Amazon 30 minute outage in 2013
 $66,240 p/m$66,240 p/m
 Approximately $3MApproximately $3M
 Its not just dollarsIts not just dollars
 Customer / end user confidenceCustomer / end user confidence
 ReputationReputation
4
Event Title
Windows Failover Cluster OverviewWindows Failover Cluster Overview
 Provides protection from unplannedProvides protection from unplanned
downtimedowntime
 Automated failoverAutomated failover
 Server-levelServer-level
 Maximum of 64 nodesMaximum of 64 nodes
 Added as a Windows Server FeatureAdded as a Windows Server Feature
5
Event Title
New Features in Windows FailoverNew Features in Windows Failover
Clustering for WS2012 & R2Clustering for WS2012 & R2
 Windows Server 2012Windows Server 2012
 64 node cluster64 node cluster
 Up from 16Up from 16
 4000 VMs per cluster4000 VMs per cluster
 1024 VMs per node1024 VMs per node
 Tunable heartbeatTunable heartbeat
 Windows Server 2012 R2Windows Server 2012 R2
 Shared virtual hard diskShared virtual hard disk
 Virtual machine drain onVirtual machine drain on
shutdownshutdown
 Virtual machine networkVirtual machine network
health detectionhealth detection
 Active Directory-detachedActive Directory-detached
clustercluster
6
Event Title
Windows Failover ClusterWindows Failover Cluster
7
Event Title
Cluster Quorum TypesCluster Quorum Types
 Node MajorityNode Majority
 For clusters with an odd number of nodesFor clusters with an odd number of nodes
 Can sustain failures of half the nodes minus oneCan sustain failures of half the nodes minus one
 Node and Disk MajorityNode and Disk Majority
 For clusters with an even number of nodesFor clusters with an even number of nodes
 Can sustain failures of half the nodes if the diskCan sustain failures of half the nodes if the disk
witness remains onlinewitness remains online
 Node and File Share MajorityNode and File Share Majority
 For clusters with special configurationsFor clusters with special configurations
 Like Node and Disk Majority but usesLike Node and Disk Majority but uses
a file share witnessa file share witness
8
Event Title
Adding the Failover ClusteringAdding the Failover Clustering
FeatureFeature
 Server ManagerServer Manager
 Add Roles andAdd Roles and
FeaturesFeatures
 FeaturesFeatures
 Failover ClusteringFailover Clustering
9
Event Title
Creating the ClusterCreating the Cluster
10
 Failover Cluster ManagerFailover Cluster Manager
 Validate theValidate the
configurationconfiguration
 Select NodesSelect Nodes
 Tests:Tests:
 SystemSystem
 NetworkingNetworking
 StorageStorage
 Create ClusterCreate Cluster
 Select NodesSelect Nodes
 Cluster nameCluster name
 Virtual IPVirtual IP
 Select QuorumSelect Quorum
 StorageStorage
Event Title
Multi-site Failover ClustersMulti-site Failover Clusters
 Disaster recoveryDisaster recovery
 Protects against loss of entire locationProtects against loss of entire location
 Storage synced with SAN-to-SANStorage synced with SAN-to-SAN
replicationreplication
11
Event Title
Multi-site Failover ClustersMulti-site Failover Clusters
 Best practicesBest practices
 Even number of nodeEven number of node
 Node and File Share Majority optionNode and File Share Majority option
 Including a file share witnessIncluding a file share witness
 Consider DNS replicationConsider DNS replication
 TTLTTL
 Tuning heartbeat to adjust for latencyTuning heartbeat to adjust for latency
 Asynchronous storage replicationAsynchronous storage replication
 Using the same HBAs and NICs on each siteUsing the same HBAs and NICs on each site
12
Event Title
Architecting AvailabilityArchitecting Availability
 TechnologyTechnology
 Failover ClusteringFailover Clustering
 PeoplePeople
 TrainingTraining
 ProcessesProcesses
 StandardsStandards
 DocumentationDocumentation
 RunbooksRunbooks
13
Event Title
FAQsFAQs
Q: Do I need specialized hardware for the failoverQ: Do I need specialized hardware for the failover
cluster nodes?cluster nodes?
A: No. You can use stand servers. Microsoft will supportA: No. You can use stand servers. Microsoft will support
the cluster if it passes the validation tests.the cluster if it passes the validation tests.
Q: Do all the cluster nodes have to be configured theQ: Do all the cluster nodes have to be configured the
same?same?
A: NoA: No
Q: Do I need a SAN if I want to buildQ: Do I need a SAN if I want to build
a Windows Failover Cluster?a Windows Failover Cluster?
A: No. But you do need shared storageA: No. But you do need shared storage
Q: can you implement a WindowsQ: can you implement a Windows
Failover Cluster on with virtual machines?Failover Cluster on with virtual machines?
A: Yes. You can implement a WindowsA: Yes. You can implement a Windows
Failover Cluster at either the hostFailover Cluster at either the host
or guest level.or guest level.
14
Event Title
Additional ReferencesAdditional References
 Failover Clustering OverviewFailover Clustering Overview
 http://technet.microsoft.com/en-us/library/hh831579.aspxhttp://technet.microsoft.com/en-us/library/hh831579.aspx
 Requirements and Recommendations for a Multi-SiteRequirements and Recommendations for a Multi-Site
Failover ClusterFailover Cluster
 http://technet.microsoft.com/en-http://technet.microsoft.com/en-
us/library/dd197575(v=ws.10).aspxus/library/dd197575(v=ws.10).aspx
 Windows Server 2012: Building a Two-Node Failover ClusterWindows Server 2012: Building a Two-Node Failover Cluster
 http://windowsitpro.com/windows-server-2012/windows-server-http://windowsitpro.com/windows-server-2012/windows-server-
2012-building-two-node-failover-cluster2012-building-two-node-failover-cluster
 Creating a Windows Server 2012 R2Creating a Windows Server 2012 R2
Failover Cluster using Starwind SAN V8Failover Cluster using Starwind SAN V8
 http://www.starwindsoftware.com/http://www.starwindsoftware.com/
white-papers/white-papers/
 Windows Server 2012 Geo-ClusteringWindows Server 2012 Geo-Clustering
 www.starwindsoftware.comwww.starwindsoftware.com
15
Event Title
mike.otey@penton.commike.otey@penton.com
@michael_otey@michael_otey
16

High Availability with Windows Server Clustering and Geo-Clustering

  • 1.
    Event Title High AvailabilitywithHigh Availability with Windows FailoverWindows Failover Clustering and Geo-Clustering and Geo- ClusteringClustering Michael Otey Senior Technical Director Windows IT Pro SQL Server Pro 1
  • 2.
    Event Title What thispresentation coversWhat this presentation covers  Today’s IT requirementsToday’s IT requirements  The costs of downtimeThe costs of downtime  Understanding Windows FailoverUnderstanding Windows Failover ClusteringClustering  New features in Ws2012 / R2 clusteringNew features in Ws2012 / R2 clustering  Geo-clustersGeo-clusters  Architecting availabilityArchitecting availability  Additional referencesAdditional references 2
  • 3.
    Event Title Today’s ITRequirementsToday’s IT Requirements  AgilityAgility  Meeting quickly changing business conditionsMeeting quickly changing business conditions  Reducing complexityReducing complexity  Increasing automationIncreasing automation  Consumerization of ITConsumerization of IT  Reducing costsReducing costs  CapEXCapEX  OpExOpEx  Data control and securityData control and security
  • 4.
    Event Title The Costsof DowntimeThe Costs of Downtime  GartnerGartner  $5600 p/m$5600 p/m  $300K p/h$300K p/h  Amazon 30 minute outage in 2013Amazon 30 minute outage in 2013  $66,240 p/m$66,240 p/m  Approximately $3MApproximately $3M  Its not just dollarsIts not just dollars  Customer / end user confidenceCustomer / end user confidence  ReputationReputation 4
  • 5.
    Event Title Windows FailoverCluster OverviewWindows Failover Cluster Overview  Provides protection from unplannedProvides protection from unplanned downtimedowntime  Automated failoverAutomated failover  Server-levelServer-level  Maximum of 64 nodesMaximum of 64 nodes  Added as a Windows Server FeatureAdded as a Windows Server Feature 5
  • 6.
    Event Title New Featuresin Windows FailoverNew Features in Windows Failover Clustering for WS2012 & R2Clustering for WS2012 & R2  Windows Server 2012Windows Server 2012  64 node cluster64 node cluster  Up from 16Up from 16  4000 VMs per cluster4000 VMs per cluster  1024 VMs per node1024 VMs per node  Tunable heartbeatTunable heartbeat  Windows Server 2012 R2Windows Server 2012 R2  Shared virtual hard diskShared virtual hard disk  Virtual machine drain onVirtual machine drain on shutdownshutdown  Virtual machine networkVirtual machine network health detectionhealth detection  Active Directory-detachedActive Directory-detached clustercluster 6
  • 7.
    Event Title Windows FailoverClusterWindows Failover Cluster 7
  • 8.
    Event Title Cluster QuorumTypesCluster Quorum Types  Node MajorityNode Majority  For clusters with an odd number of nodesFor clusters with an odd number of nodes  Can sustain failures of half the nodes minus oneCan sustain failures of half the nodes minus one  Node and Disk MajorityNode and Disk Majority  For clusters with an even number of nodesFor clusters with an even number of nodes  Can sustain failures of half the nodes if the diskCan sustain failures of half the nodes if the disk witness remains onlinewitness remains online  Node and File Share MajorityNode and File Share Majority  For clusters with special configurationsFor clusters with special configurations  Like Node and Disk Majority but usesLike Node and Disk Majority but uses a file share witnessa file share witness 8
  • 9.
    Event Title Adding theFailover ClusteringAdding the Failover Clustering FeatureFeature  Server ManagerServer Manager  Add Roles andAdd Roles and FeaturesFeatures  FeaturesFeatures  Failover ClusteringFailover Clustering 9
  • 10.
    Event Title Creating theClusterCreating the Cluster 10  Failover Cluster ManagerFailover Cluster Manager  Validate theValidate the configurationconfiguration  Select NodesSelect Nodes  Tests:Tests:  SystemSystem  NetworkingNetworking  StorageStorage  Create ClusterCreate Cluster  Select NodesSelect Nodes  Cluster nameCluster name  Virtual IPVirtual IP  Select QuorumSelect Quorum  StorageStorage
  • 11.
    Event Title Multi-site FailoverClustersMulti-site Failover Clusters  Disaster recoveryDisaster recovery  Protects against loss of entire locationProtects against loss of entire location  Storage synced with SAN-to-SANStorage synced with SAN-to-SAN replicationreplication 11
  • 12.
    Event Title Multi-site FailoverClustersMulti-site Failover Clusters  Best practicesBest practices  Even number of nodeEven number of node  Node and File Share Majority optionNode and File Share Majority option  Including a file share witnessIncluding a file share witness  Consider DNS replicationConsider DNS replication  TTLTTL  Tuning heartbeat to adjust for latencyTuning heartbeat to adjust for latency  Asynchronous storage replicationAsynchronous storage replication  Using the same HBAs and NICs on each siteUsing the same HBAs and NICs on each site 12
  • 13.
    Event Title Architecting AvailabilityArchitectingAvailability  TechnologyTechnology  Failover ClusteringFailover Clustering  PeoplePeople  TrainingTraining  ProcessesProcesses  StandardsStandards  DocumentationDocumentation  RunbooksRunbooks 13
  • 14.
    Event Title FAQsFAQs Q: DoI need specialized hardware for the failoverQ: Do I need specialized hardware for the failover cluster nodes?cluster nodes? A: No. You can use stand servers. Microsoft will supportA: No. You can use stand servers. Microsoft will support the cluster if it passes the validation tests.the cluster if it passes the validation tests. Q: Do all the cluster nodes have to be configured theQ: Do all the cluster nodes have to be configured the same?same? A: NoA: No Q: Do I need a SAN if I want to buildQ: Do I need a SAN if I want to build a Windows Failover Cluster?a Windows Failover Cluster? A: No. But you do need shared storageA: No. But you do need shared storage Q: can you implement a WindowsQ: can you implement a Windows Failover Cluster on with virtual machines?Failover Cluster on with virtual machines? A: Yes. You can implement a WindowsA: Yes. You can implement a Windows Failover Cluster at either the hostFailover Cluster at either the host or guest level.or guest level. 14
  • 15.
    Event Title Additional ReferencesAdditionalReferences  Failover Clustering OverviewFailover Clustering Overview  http://technet.microsoft.com/en-us/library/hh831579.aspxhttp://technet.microsoft.com/en-us/library/hh831579.aspx  Requirements and Recommendations for a Multi-SiteRequirements and Recommendations for a Multi-Site Failover ClusterFailover Cluster  http://technet.microsoft.com/en-http://technet.microsoft.com/en- us/library/dd197575(v=ws.10).aspxus/library/dd197575(v=ws.10).aspx  Windows Server 2012: Building a Two-Node Failover ClusterWindows Server 2012: Building a Two-Node Failover Cluster  http://windowsitpro.com/windows-server-2012/windows-server-http://windowsitpro.com/windows-server-2012/windows-server- 2012-building-two-node-failover-cluster2012-building-two-node-failover-cluster  Creating a Windows Server 2012 R2Creating a Windows Server 2012 R2 Failover Cluster using Starwind SAN V8Failover Cluster using Starwind SAN V8  http://www.starwindsoftware.com/http://www.starwindsoftware.com/ white-papers/white-papers/  Windows Server 2012 Geo-ClusteringWindows Server 2012 Geo-Clustering  www.starwindsoftware.comwww.starwindsoftware.com 15
  • 16.