Snapshot Replication and Synchronous Replication
HPE Nimble Storage supports two types of replication, snapshot replication and synchronous replication. Snapshot replication creates a backup version of your volumes so that you have a point-in-time copy to refer back to if a prior version of the data is needed. Synchronous replication provides a replica of the volumes that is created synchronously with no delay in the recovery point. It is intended for immediate recovery due to a site failure.
To assist in understanding the differences between snapshot replication and synchronous replication, the following table presents a comparison of the two.
Behavior | Snapshot Replication | Synchronous Replication | Notes |
---|---|---|---|
Recovery Point Objective (RPO) | Depending on the replication snapshot schedule. If the schedule is 15 minutes, then the RPO is 15 minutes | 0 RPO Note: RTO can be < 1 minute when in a single site or when metro stretch
clustering is not being used.
|
Failures caused by incorrect writes are not covered by synchronous replication. They can be recovered from snapshots. |
Automatic recovery | No | Yes, if Automatic Switchover is configured | Only available with synchronous replication when accompanied by the Automatic Switchover (ASO) feature. |
Manual recovery | Yes | Yes | |
Replication Partners | Group partners | Pool partners | Pool partners are auto-generated and deleted when the pool is deleted or merged |
Maximum number of partners | 2 | 1 | |
Number of protected volumes | 1024 volumes | 128 volumes upstream and 128 volumes downstream | |
Replication configuration level | Volume collection | Volume collection | Synchronous replication is not allowed on multi-pool volume collections |
Maximum number of protection schedules per volume collection | 10 | 10 | |
Extent of replication | The maximum supported number of groups is four | Two arrays, two single-array pools | |
Replication type | Between groups | Within group | |
Replicate sub-set of Snapshots | Yes | Yes | |
Hardware array models supported | All models | AFxx, HFxx, AFxxxx and CSxxxx arrays only | For synchronous replication, the upstream array must be the same model as the downstream array |
Synchronous mode | N/A | Soft synchronous mode where when a volume goes out-of-sync, it continues to accept writes | |
Resume automatically | Yes | Yes, through an automatic resynchronization | |
Site fault tolerance | Yes | Yes | |
Recovery time objective (RTO) | User initiated handover can be completed in under one minute | Switchover can be completed in under one minute | |
Network connectivity for replication traffic between partners | IP | IP | |
Support for multiprotocol arrays (Fibre Channel and iSCSI) | Yes | No | |
Link bandwidth | N/A | 10 Gpbs | |
Number of links | At least one | At least two links are recommended for redundancy and performance | |
Add support non-disruptively | Yes | Yes | |
Remove support non-disruptively | Yes | Yes | |
Array replacement allowed | Yes | Yes, see note |
Note: To replace an array, un-configure synchronous replication, move
data to a new array, then reconfigure synchronous
replication.
|
Changing replication type allowed | No | No | |
Volume resize support | Yes | Yes, see note |
Note: To resize a volume with synchronous replication, unconfigure
synchronous replication, resize the volume, then reconfigure
synchronous replication.
|
Subnets | Arrays do not need to be in the same subnet | All arrays must be in the same subnet | |
Protection templates | Yes | None provided | |
Location of upstream and downstream replicated volumes | In separate groups, so in separate volume collections. | All in one volume collection | |
Location of snapshots | In separate groups, so in separate snapshot collections. | Two separate snapshot collections, one for upstream and one for downstream | |
Host reconfiguration required after handover | Yes | No |