6.3 Unclustered Server Pools

When creating a server pool, you specify whether the servers in the pool will be part of a cluster or not. In most cases, you create a clustered server pool. You can create a non-clustered pool when all servers in the pool are expected to use only NFS shares as repositories. If your Oracle VM Servers are also expected to access repositories on physical disks, then these servers must be part of a clustered server pool.

Figure 6.2, “Unclustered Server Pools Using Only NFS Storage” illustrates server pools in an unclustered configuration, with shared access to resources on NFS storage but no HA features for the servers.

Figure 6.2 Unclustered Server Pools Using Only NFS Storage

This figure shows an Oracle VM configuration with two unclustered server pools. They have access to shared attached storage on NFS servers and also share a storage repository between them, for example as a common location for templates and ISOs. A server pool file system is not used for unclustered server pools.

Non-clustered server pools do not require a server pool file system, though a Virtual IP is still required and the Master function is also assigned to one of the server pool members.

A non-clustered server pool does not support HA for virtual machines deployed on its servers. If a server fails, the virtual machines on this server have to be restarted manually on a server in this server pool, or possibly on a server in another server pool, if that server pool also has access to the repositories needed for deploying the virtual machines on the failed server. Live Migration is supported between servers in a non-clustered pool if the servers have the same CPU affinity (same family and type of CPU).


Converting non-clustered server pools to clustered server pools is not supported in Release 3.2 of Oracle VM.