High availability (HA) cluster configuration

A cluster consists of a group of servers that work together for high availability of the hypervisor. Using a cluster-type architecture lets you virtually separate redundant pairs, and supports larger, more complex systems.

For example, within vSphere, clustering adds the capability to automatically manage the distribution of VMs across hosts via the Distributed Resource Scheduler (DRS), as well as allowing the use of vMotion to migrate VMs between hosts. Thus, clustering enhances High Availability (HA), resource sharing, load-balancing, and makes management of the hosted VMs easier. However, in a control system environment where VMs may be sensitive to data changes during runtime, you should exercise extra care to ensure that your cluster is configured correctly.

High availability and load balancing leverage features like VMware vSphere HA, Microsoft Failover Clustering, along with the native application-level redundancy provided by System Platform.

HA configurations are highly recommended for critical infrastructure and other applications where uptime is essential.

As is the case with using multiple host servers, it is a best practice to place redundant pairs on different host servers (hypervisors).

Examples of vSphere configuration to keep machines separate or together:

In vSphere, you can specify which VMs should be kept on the same host and which ones should be kept on separate hosts, based on the role of the VM.

Planning and management may be simplified if you specify that all primary and backup VMs are kept on separate hosts.

Example: Keeping machines together

It is recommended that you associate System Platform VMs to high performing hypervisors in the cluster.

You can choose to pin virtual machines to certain hosts. Keep machines together ensures that these VMs are placed on the same host whenever possible. Leverage naming conventions to ensure that nodes are mapped to the correct host.

Example: Keeping machines separate

It is recommended that you associate redundant System Platform VMs to separate hypervisors in the cluster.

The option "Keeping virtual machines separate" ensures that selected VMs are never on the same host.