These notes are my personal notes from the FREE training on Pluralsight. You can get your FREE signup through technet/MSDN or Dreamspark. The title of this course is exactly the title of this post. These notes are from this specific course only. I use these as a refresher Study Guide. POWERSHELL topics andcmdlets are in purple. I have a few notes with the “DEMO” each time the training included a DEMO just so you can see how many demos there were which were really helpful. Thanks to Greg Shields @ConcentratdGreg, the trainer, contact info at the end.

All, or nearly all, sections include DEMOS so I’m not notating that separately.

1. Configure Network Load Balancing
   1. most commonly used with IIS
   2. stateless (doesn’t matter what node user connects with)
   3. Configure NLB Prerequisites
      1. install Feature
      2. Unicast, Multicast, IGMP Multicast
      3. Unicast
         1. always works
         2. 1:1
         3. requires a second NIC on each server
         4. causes subnet flooding; all traffic to all hosts goes to all hosts
      4. Multicast
         1. no second NIC
         2. network configurations
         3. does not solve subnet flooding
      5. IGMP Multicast – best practice
         1. no second nic
         2. network requirements
         3. solve subnet flooding problem
   4. Install NLB Nodes
   5. Configure Cluster Operation Mode
   6. Configure Port Rules and Affinity
   7. Upgrade an NLB Cluster
2. Configure Failover Clustering (read prior post here)
   1. Cluster Storage
      1. shared storage is not built in Windows; it’s a foreign concept
      2. proper configuration of storage is critical
      3. iSCSI, FC, Storage Spaces (in our previous FS training)
      4. we’re using iSCSI here in this demo
   2. Configure Cluster networking
      1. best practice to separate cluster private network and storage network
      2. Failover Cluster Manager – console for cluster management
      3. Cluster Validation wizard (lots of experience with this 😉
      4. In this Demo, Cluster Private network, Storage network, and Management / Production
      5. Check the networks in Failover Cluster Manager
   3. Cluster Shared Volumes (CSV) used by Hyper-V virtual machines
      1. Quorum (chosen by smallest size)
      2. Available Storage LUNs (if containing a VM, they all would have to fail over at the same time (or each have dedicated LUN)
      3. CSV, each VM can fail over individually
      4. you can define a disk as a CSV, and you can revert also.
      5. More here on Using CSV for Failover Cluster
         1. CSV cache size configuration; (Get-Cluster).BlockCacheSize = 512 for Server 2012 R2, for more read the link above.
      6. Quorum configuration
         1. Quorum is only to identify if enough of a cluster remains to still operate as a quorum.
         2. Quorum Models (dependent on number of nodes)
            1. Node majority (used for ODD number of nodes)
            2. Node and Disk Majority (even number of nodes)
            3. “split brain” when cluster breaks into two separate groups of nodes who each think they are quorum
            4. No Majority Disk Only (old, not used any more)
            5. Node and File Server Majority (special considerations) even nodes, multi site. The separate vote goes to a disk file share somewhere.
         3. Configure quorum model in Failover Cluster Manager
            1. MS automatically manages cluster quorum setting now. “use default quorum configuration”
            2. Or, you can go to advanced features and dance with the complexity on your own.
      7. Clusters without network names (detached cluster) Deploy an Active Directory-Detached Cluster
         1. SQL server outside your firewall
         2. not supported for several additional roles
         3. no bitlocker
         4. no CAU (Cluster Aware Updating)
         5. read the link
         6. cannot use FOCM
         7. PS – (Get-Cluster).AdministrativeAccessPoint
            1. read the link for more
      8. CAU (Cluster Aware Updating)
         1. “update cluster” item in Server Manager
         2. allows cluster to manage resource movement to update nodes/hosts.
         3. configure self-updating options wizard
         4. add the ROLE on the cluster
         5. choose schedule (normal WSUS stuff)
         6. reboot timeouts, max retries, pre or post scripts, recommended / important
         7. All the above is for Windows updates, not WSUS
         8. “Analyze cluster updating readiness”
         9. PS
            1. Cluster-Aware Updating Cmdlets in Windows PowerShell
      9. Restoring single node of cluster
         1. Evict = kicking node out
         2. restore configuration from backup (make sure you have system state)
      10. Upgrading a cluster
          1. not recommended to directly upgrade a cluster
          2. this is a cut and move
          3. “copy cluster roles” from a wizard from the TARGET cluster, connect to OLD cluster to get configs.
3. Manage Failover Clustering Roles
   1. remember MSCS is a “general purpose” clustering solution
   2. role-specific settings
      1. DFS, SHCP, DTC, FIle Server, iSCSI target, etc., etc..
      2. Generic application, script, or service
      3. DEMO – clustered NOTEPAD via Generic Application
      4. cluster is a SINGLE instance of the app that fails from node to node, moving the resources (including created drives) as needed
      5. Continuously available file server
         1. General Use, or SOFS (Scale Out File Server) (used for Hyper-V and SQL)
      6. Configure Virtual Machines
         1. do not put SOFS and VMs on same CSV
   3. fail-over and preferences
      1. ROLES (shared app, file server, VM, etc.)
      2. move, stop, change startup priority
      3. no autostart
      4. add resources or storage
      5. “show dependency” report
         1. graphical representation of dependencies
      6. “preferred owner” unchecked can be used, just not preferred
      7. failover max
      8. failback now/yes, set hours it can happen
      9. cluster handles DNS records for cluster required records
      10. you can manually add dependencies
   4. possible and preferred owners
      1. possible owners (cannot be on any node that is not checked)
      2. preferred owner (can use unchecked nodes, they’re just not preferred)
   5. guest clustering
      1. another layer of abstraction
      2. simply means clustering VMs that are on the MSCS cluster
      3. shared .VHDX
      4. new feature (like RDMs) in VMware
      5. advanced features “enable virtual hard disk sharing”
4. Manage VM Movement
   1. Migration – Live, Quick, Storage
      1. Quick
         1. the old fashioned, with a quick period of loss of service
      2. Live
         1. no loss of service
      3. Storage
         1. moving the .vhdx, the data
      4. Quick is technically faster, and uses less bandwidth than Live
      5. Live – procs need to be same manufacturer and similar family
      6. virtual switches needs to be named the same
      7. physical devices must be disconnected
      8. DEMO
         1. constrained delegation has to be configured to the hosts that you want to migrate to/from
         2. CredSSP alternative to Kerbos/constrained delegation but CredSSP requires you to log onto the machine to start the migration
   2. Import, Export, Copy
      1. have to export/import if you can’t do quick/live migration
   3. Configure VM Health Protection
      1. move to locations without proper networks, or something similar
      2. VMHP is under Network Adapter / advanced features
      3. it will move it back to prior location if it ends up isolated
      4. ENABLED by default
   4. Configure Drain on Shutdown
      1. drain a node on shutdown
      2. ENABLED by default
   5. Configure VM Monitoring
      1. “resources” tab on bottome of Failover Cluster Manager
      2. checkbox to enable automatic recovery for application health monitoring
      3. if/when enabled, you can select services via checkbox that you want to include for application monitoring.