Configuring disaster recovery using MetroCluster

Definition of terms

This appendix uses the following terms when describing disaster recovery using filers in a MetroCluster configuration:

Cluster--Two filers that are connected in a cluster configuration
Cluster node--One of the filers in the cluster
Local cluster node--The cluster node in the same area as the system administrator
Remote cluster node--The cluster node in the area affected by the disaster
Remote storage--The storage that is accessible to the local cluster node, but is located at the remote cluster node site
Site--The area in which a cluster node is located, for example, a building

Mirrored volume configuration

The MetroCluster configuration employs SyncMirror and a cluster configured for mirrored volumes to build a system that provides data even after complete loss of one of the cluster nodes and the storage at that site. Mirrored volumes are configured as follows:

The root volume must be mirrored with one plex at each site.
Data volumes should be mirrored with one plex at each site.
Plexes local to the filer at each site must be connected to the disk pool named pool0. Pool0 consists of the storage attached to host adapters in slots 3 through 7. See the System Configuration Guide at now.netapp.com and the Cluster Guide for your filers for more information about filer slot assignments.
Caution
The MetroCluster configuration can preserve data only if volumes are mirrored. Unmirrored volumes are not accessible if a disaster occurs; data can be lost.

MetroCluster configurations

There are two MetroCluster configurations. One configuration directly connects the two cluster nodes to each other. The other configuration uses Brocade switches to connect the two cluster nodes. The configuration you use depends on the distance between the two cluster nodes. The following sections discuss the two MetroCluster configurations: