Nov
15

Acropolis – Storage High Availability Path Selection

With an Acropolis cluster, a rerouting decision is taken on a per-vDisk basis. When the local Controller VM becomes unavailable, iSCSI connections are individually redirected to a randomly selected healthy Controller VM, resulting in distribution of load across the cluster.

Instead of maintaining live, redundant connections to other Controller VMs, the Acropolis hypervisor initiates an iSCSI connection to a healthy CVM only when the connection is required. When the local Controller VM becomes available, connections to other Controller VMs are terminated and the guest VMs reconnect to the local Controller VM.

This helps the platform maintain consistent high peformance even in the face of a failure or maintenance operation.

Speak Your Mind

*