When you create the cluster nodes that make up vRealize Operations, additional best practices improve performance and reliability in vRealize Operations.
- Deploy vRealize Operations analytics cluster nodes in the same vSphere cluster in a single data center and add only one node at a time to a cluster allowing it to complete before adding another node.
- If you deploy analytics cluster nodes in a highly consolidated vSphere cluster, you might need resource reservations for optimal performance.
Determine whether the virtual to physical CPU ratio is affecting performance by reviewing CPU ready time and co-stop.
- Deploy analytics cluster nodes on the same type of storage tier.
- To continue to meet analytics cluster node size and performance requirements, apply storage DRS anti-affinity rules so that nodes are on separate datastores.
- To prevent unintentional migration of nodes, set storage DRS to manual.
- To ensure balanced performance from analytics cluster nodes, use ESXi hosts with the same processor frequencies. Mixed frequencies and physical core counts might affect analytics cluster performance.
- To avoid a performance decrease, vRealize Operations analytics cluster nodes need guaranteed resources when running at scale. The vRealize Operations Knowledge Base includes sizing spreadsheets that calculate resources based on the number of objects and metrics that you expect to monitor, use of HA, and so on. When sizing, it is better to over-allocate than under-allocate resources.
- Because nodes might change roles, avoid machine names such as Primary, Data, Replica, and so on. Examples of changed roles might include making a data node into a replica for HA, or having a replica take over the primary node role.
- The NUMA placement is removed in the vRealize Operations 6.3 and later. Procedures related to NUMA settings from the OVA file follow:
Table 1. NUMA Setting Action Description Set the vRealize Operations cluster status to offline
- Shut down the vRealize Operations cluster.
- Right-click the cluster and click .
- Click Configuration Parameters. In the vSphere Client, repeat these steps for each VM.
Remove the NUMA setting
Note: To ensure the availability of adequate resources and continued product performance, monitor vRealize Operations performance by checking its CPU usage, CPU ready and CPU contention time.
- From the Configuration Parameters, remove the setting numa.vcpu.preferHT and click OK.
- Click OK.
- Repeat these steps for all the VMs in the vRealize Operations cluster.
- Power on the cluster.