Considerations for a fast failover configuration
For VCS service groups that contain many disk groups, you can greatly reduce failover time by implementing fast failover.
Fast failover speeds up the failover of storage resources in several ways:
-
Fast failover provides a "read-only deported" mode for disk groups on inactive nodes. This mode speeds up the process of importing a disk group.
-
Fast failover maintains the current disk group configuration in memory on the inactive nodes. Any changes are automatically synchronized so that all nodes maintain an identical disk group configuration.
For more details about fast failover, refer to the Storage Foundation Administrator's Guide.
Take the following storage-related requirements into account if you are planning to implement fast failover:
-
Fast failover is currently not supported for the following:
-
RAID-5 volumes
-
SCSI-2
-
Active/Passive (A/P) arrays for DMP
-
-
In synchronous mode of replication, if fast failover is set, then the RVG cannot be stopped and started when a disk group fails over to another node. If the RLINK is in hard synchronous mode, it may not be connected when the volume arrives, and the I/Os may fail. In such case, the Event Viewer displays NTFS and ReFS errors and file system reports the volume as RAW. Therefore, fast failover is not supported if the RLINK is in hard synchronous mode.
-
The disk group version must be 60 or later for fast failover to work. To verify the disk group version, from the VEA console, right-click the disk group and click Properties. Disk group version upgrade is required after upgrading SFW HA on the cluster nodes. Refer to the Arctera InfoScale® Installation and Upgrade Guide for more information.
Related information