I’m very often asked or have seen confusion around the configuration option’s and possibilities for an “load balancing control mechanism” for highly available virtual machines within a failover cluster environment.
a.e. which VMs should run “preferred” (-> preferred owner) on dedicated node(s) or at the other side you want to control which VMs should never run together at 1 node (-> AntiAffinityClassNames).
a.e. from node/host capacity planning perspective you’re also able to configure on which node(s) your VM(s) can be failed over (-> Possible Owners)
In combination all of these parameters (AntiAffinityClassNames, Possible Owner, Preferred Owner….) you can configure a very well “balanced” environment to get the best performance for your VMs/Applications and a well utilized server hardware.
Interested? here we go::
Failover Cluster – WMI MSCluster_ResourceGroup Class
How to configure Windows clustering groups for hot spare support
Failover behavior on clusters of three or more nodes
Windows Server 2008 R2 Failover Cluster PowerShell Command Reference
Controlling Server Workload in MSCS Clusters (2003)
In addition Elden Christensen (Senior PM for Cluster, Microsoft) has written an great blog around failover policies for Virtual Machines in Failover Cluster environments:
Understanding Hyper-V Virtual Machine (VM) Failover Policies
Any further questions? let me know….
Stay tuned for further blogs around failover policies