See also Data Protection Glossary of Terms.
Asynchronous replication. The process of replicating data from a protected path on one cluster to other clusters, by means of periodic snapshots, taken of the path on the source cluster, and copied to the other destination clusters. These snapshots can be used in the event the path on the source cluster is not longer available.
Destination peer. A replication peer with role destination, which means maintains copies of data in the source peer. For asynchronous replication, these copies are snapshots of the protected path on the source; for synchronous replication, the destination is fully synchronized with the source path.
Failback. The process of restoring a group of replication peers to their original state, after a failure in the primary peer is corrected. It typically involves the original primary peer, now restored, resuming the role of source, and the original destination peer, which was made the source during the failover, resuming the role of destination.
Failover. The process of switching the source for a protected path to the copy on a destination replication cluster. This is typically done when the path on the source cluster is no longer available.
Global synchronization. The synchronization of an NFSv3 view on one cluster with views on other clusters that allows for seamless failover from one cluster to another in the event of a failure. With seamless failover, clients continue to access the view using the same mount point, even if the cluster has failed, and they are in fact accessing synchronized data from another cluster. See Seamless Failover.
Graceful Failover. In a graceful failover no data would be lost, since the destination would be synced with the source before the destination becomes writeable. However, the failover itself will take longer the more delta there is to sync since the last restore point completed at the destination.
Alternative text:The process of making the destination peer the source peer for the protected path, without any loss of data. For asynchronous replication, this involves first synchronizing the destination peer with the source peer from the time of the last snapshot. For synchronous replication the destination is always fully synchronized to the source.
Path on peer. The path on a async replication peer to which data is replicated. This can be a different path to the path protected on the source peer. When first configured, it must be a path that does not yet exist on the destination peer.
RPO. Recovery Point Objective. The age of the latest data replicated to an async replication peer, per protected path. It is measured as the time since the snapshot that was last completely replicated to the destination peer was created on the source peer. This is not relevant for synchronous replication.
RPO can be viewed in the predefined Analytics report RPO for the object type Protected Path.
Note
Although RPO can also refer to the maximum tolerated age of the latest replicated data, it is not used in this sense in VAST Cluster interfaces.
RPO Exception. A situation in which a snapshot is skipped because replication to the destination peer exceeded the interval between snapshots. This typically occurs during initial sync and can also occur if the replication interval is set too low for the data workload and the replication connection.
RPO Exception can be viewed per protected path in the predefined Analytics report RPO Miss for the object type Protected Path.
Seamless failover. The ability of NFSv3 clients to continue running applications against a mounted view (on one cluster), without the need to remount the view when the cluster fails over to a replication peer. See Global Synchronization.
Source peer. A replication peer with the role source. For asynchronous replication, snapshots are taken periodically and replicated from the source replication peer to one or more destination replication peer(s). For synchronous replication, data is continuously synchronized with a single destination replication peer.
Standalone peer. A async replication peer with the standalone role which reflects that replication from the source peer to the (standalone) peer is suspended and the peer is writeable.
Synchronous replication. The process of replicating data from a protected path on one (source) cluster to another (destination) cluster, maintaining both clusters fully synchronized to each other. Initially, the source path is copied to the destination peer. Then, all write operations on either the source or destination peers are immediately synchronized to the other cluster.
Ungraceful Failover. In an ungraceful failover event, you recover only data that was saved in the last restore point that was completed on the destination peer. Any data that was written to the primary cluster since the data captured in that restore point would be lost.