Configuration variations for fabric-attached MetroCluster configurations

Fabric-attached MetroCluster configurations support asymmetrical and active/passive configurations.

The following list describes some common configuration variations that are supported for fabric-attached MetroCluster configurations:

  • Asymmetrical mirroring

    You can add storage to one or both nodes that is not mirrored by the other node. However, any data contained in the unmirrored storage could be lost if that site experiences a disaster.

    • Multiple disk failures in an unmirrored aggregate (three or more disk failures in a RAID-DP aggregate, two or more disk failures in a RAID4 aggregate) will cause the node to panic, resulting in a temporary data service outage while the node reboots or disaster recovery is performed. You must mirror the root volumes to enable successful takeover.
    • If the system contains unmirrored aggregates using disks from remote site, issues in ISLs can lead to multiple disk failure. Therefore, if you see ISL instability, such as ISL going down, switch failures and so on, you must take such unmirrored aggregates offline.
    • You must connect the unmirrored storage to both nodes, just as for mirrored storage. You cannot have storage that is connected to only one node in an HA configuration.
    • If the ISL links are being taken down for maintenance and your system contains unmirrored aggregates, you must first take all unmirrored aggregates offline.
  • Active/passive MetroCluster configurations

    In this configuration, the remote (passive) node does not serve data unless it has taken over for the local (active) node. Mirroring the passive node’s root volume is optional. However, both nodes must have all licenses for MetroCluster configuration installed so that remote takeover is possible.