Replication Partners and How Replication Works

You can set up replication between two Nimble arrays or a Nimble array and HPE Cloud Volumes (HPE CV). By creating a partner on each of the two arrays, replication is automatically set up. You can configure Nimble arrays to have up to fifty replication partners (also known as replication stores in HPE CV). Replication is automatic, based on the protection schedule assigned to the volume or volume collection.

Replication partners can be reciprocal, upstream (the source of replicas), or downstream (the receiver of replicas) partners. You can have several upstream arrays that replicate to one downstream partner.

When you create a replication partner, you enable one Nimble array to perform replication tasks on another array. The two arrays must be able to communicate over a network.

Note:

Replication uses ports 4213 and 4214. Replication of encrypted volumes uses port 5391. See Configure Firewall Ports for more information.
Replicating encrypted volumes to HPE CV requires NimbleOS 5.0.6.x or higher.
Replication to HPE CV is supported only on iSCSI arrays. Replication with Nimble arrays can use different access protocols. One of the partners can be an iSCSI array and its replication partner can be a Fibre Channel array.

Replication partners can use different access protocols. One of the partners can be an iSCSI array and its replication partner can be a Fibre Channel array.

Note: If you create volume collections that use replication, configure both arrays as replication partners before you begin.

Volumes are replicated by associating them with volume collections with at least one replication schedule. A volume collection can have up to ten protection schedules. One or more of these protection schedules can be replicating schedules. All replicating schedules of a volume collection are replicated to the same replication partner.

Using volume collection schedules, you define which replication partner and at what times volumes should be replicated, and how many point-in-time versions of the replica should be retained. You can assign volumes to the necessary volume collections. At the scheduled time, the array from which the volume is being replicated sends the data to the downstream (receiving) replication partner. You can also create a new protected standalone volume that is automatically associated with a volume collection instead of creating a volume first and then associating it with a volume collection.

If a volume uses a custom performance policy, you must duplicate that policy on the replication partner.

When you enable replication of a volume for the first time, the entire data set on the volume at the time of the snapshot is replicated from the original array group (upstream) to the destination replication partner (the downstream array group). Subsequent updates replicate the differences between the last replicated snapshot on the upstream partner to the downstream partner.

It is recommended that you configure the snapshot retention count on the replica (downstream) array group to at least the number of snapshots on the upstream array group, matching the frequency of snapshots.

Note: In majority of NimbleOS releases, by default, the replication partner retains only two snapshots for each replica volume in the volume collection if configuration is not adjusted.

Snapshot collections are replicated in the order that the collections were taken. Once replication is caught up, the upstream and downstream replication partners only retain as many snapshots as set in the retention criteria. The system deletes pending snapshot collections that exceed the retention criteria.