13.4.2. Setting a Cluster Resilience Policy

A cluster resilience policy determines the action to be taken for the virtual machines running on a host which has shut down unexpectedly or been put into maintenance.
The next available host depends on the cluster policy you have set, as described in Section 12.3.2, “Cluster Policy”. For example, if your cluster policy is set to Even Distribution, the next available host refers to the host which is running the lightest CPU workload at the time migration is triggered. The exception to this rule is the host which is elected as the Storage Pool Manager, because more of its CPU resources are in use for writing metadata from all virtual machines in the system. Therefore, the SPM tends to run less virtual machines than other hosts.

To define a cluster resilience policy:

  1. On the Clusters tab which displays on the results list, select the cluster and click Edit. The Edit Cluster dialog displays. Select the Resilience Policy tab.
    Cluster Resilience Policy

    Figure 13.5. Cluster Resilience Policy


  2. Select the policy to use. The available options are:
    • Migrate Virtual Machines: This option migrates all virtual machines when a host becomes non-operational. The virtual machines are migrated in order of their defined priority, so the highly available virtual machines are at the front of the migration queue.
    • Migrate only Highly Available Virtual Machines: This option migrates only the highly available virtual machines, so the other available hosts in the cluster do not become overloaded if they are to rescue virtual machines from non-operational hosts.
    • Do Not Migrate Virtual Machines: This option prevents your virtual machines from being migrated at all.

    Note

    Virtual machine migration is a network-intensive operation. For instance, on a setup where a host is running ten or more virtual machines, migrating all of them can be a long and resource-consuming process. Therefore, select the policy action to best suit your setup. If you prefer a conservative approach, disable all migration of virtual machines. Alternatively, if you have many virtual machines, but only several which are running critical workloads, select the option to migrate only highly available virtual machines.
  3. Click OK to save the policy selection for virtual machines in the cluster.