Clusters and server farms
The key differences between a cluster and a server farm are the behavior when a failure occurs, ease of management, and scalability of operations:
- A cluster allows an EVS to be automatically migrated among cluster nodes in the event of a failure, and management of all nodes in the cluster is centralized. The use of a single namespace allows clients to mount a single network resource, while having the actual storage virtualized among different devices in the cluster.
- A server farm allows an EVS to be migrated manually among servers in the server farm, but this is a manual process, and it does not happen automatically in the event of a failure. All servers in the server farm may be managed by a single NAS Manager, but each server must be managed as an independent unit.
Clusters versus server farms
The following table distinguishes the properties of a cluster and a server farm:
Property | Cluster | Server farm |
Can belong to a server farm | Yes | No |
EVS migration under server failure | Automatic | Manual |
NVRAM mirroring between servers | Yes | No |
Maximum number of storage servers | Depends on several factors. See Maximum number of nodes supported for more information. | No explicit restriction on the number of servers; however, a NAS Manager can manage only eight quorum devices and server farm planning should be adjusted accordingly. |
Shared NAS Manager | For central management; cluster quorum | For central management; EVS migration |
Storage Pools | Yes | No |
Common Storage Access | Yes | Yes |