VMware DRS is a load-balancing feature that reassigns VMs to other VMs in a cluster of ESX hosts, to balance the workload on hosts across the cluster. Using the vMotion network link, DRS migrates VMs in accordance to host load requirements, VM load requirements, and user demands. Shared storage is required in order to use DRS.
DRS can be configured to do load balancing automatically, or to make recommendations, which can then be implemented manually. Recommendations are made by the vCenter Server in accordance to a user-selectable migration threshold scale of predicted performance improvements, ranging from conservative to aggressive.
Because IP Availability Manager typically has a large memory footprint and variable CPU requirements that range from a high rate of CPU consumption (during network discovery and codebook computation) to a lower, steady rate of CPU consumption (during periods of monitoring and polling), take care when choosing a migration threshold for DRS. Choosing a migration threshold that is too aggressive could cause the IP Availability Manager VM to migrate when its CPU usage suddenly increases.
Be conservative with DRS migration thresholds, or enable other VMs to be migrated rather than the IP Availability Manager VM.
DRS is supported by IP Availability Manager, but not recommended because of its large, active real memory working set. Migration requires that resources be available simultaneously on the source and destination host, and in the case of IP Availability Manager, these resources, especially memory, could be significant, depending on the size of the managed topology.
