SnapMirror unified replication basics
SnapMirror unified replication allows you to configure disaster recovery and archiving on the same destination volume. When unified replication is appropriate, it offers benefits in reducing the amount of secondary storage you need, limiting the number of baseline transfers, and decreasing network traffic.
How unified data protection relationships are initialized
As with SnapMirror, unified data protection performs a baseline transfer the first time you invoke it. The SnapMirror policy for the relationship defines the contents of the baseline and any updates.
A baseline transfer under the default unified data protection policy MirrorAndVault
makes a Snapshot copy of the source volume, then transfers that copy and the data blocks it references to the destination volume. Like vault archiving, unified data protection does not include older Snapshot copies in the baseline.
How unified data protection relationships are updated
At each update under the MirrorAndVault
policy, SnapMirror creates a Snapshot copy of the source volume and transfers that Snapshot copy and any Snapshot copies that have been made since the last update, provided they have labels matching the labels defined in the Snapshot policy rules. In the following output from the snapmirror policy show
command for the MirrorAndVault
policy, note the following:
-
Create Snapshot
is “true”, indicating thatMirrorAndVault
creates a Snapshot copy when SnapMirror updates the relationship. -
MirrorAndVault
has rules “sm_created”, “daily”, and “weekly”, indicating that both the Snapshot copy created by SnapMirror and the Snapshot copies with matching labels on the source are transferred when SnapMirror updates the relationship.
cluster_dst::> snapmirror policy show -policy MirrorAndVault -instance Vserver: vs0 SnapMirror Policy Name: MirrorAndVault SnapMirror Policy Type: mirror-vault Policy Owner: cluster-admin Tries Limit: 8 Transfer Priority: normal Ignore accesstime Enabled: false Transfer Restartability: always Network Compression Enabled: false Create Snapshot: true Comment: A unified SnapMirror synchronous and SnapVault policy for mirroring the latest file system and daily and weekly snapshots. Total Number of Rules: 3 Total Keep: 59 Rules: SnapMirror Label Keep Preserve Warn Schedule Prefix ---------------- ---- -------- ---- -------- ------ sm_created 1 false 0 - - daily 7 false 0 - - weekly 52 false 0 - -
Unified7year policy
The preconfigured Unified7year
policy works exactly the same way as MirrorAndVault
, except that a fourth rule transfers monthly Snapshot copies and retains them for seven years.
Rules: SnapMirror Label Keep Preserve Warn Schedule Prefix ---------------- ---- -------- ---- -------- ------ sm_created 1 false 0 - - daily 7 false 0 - - weekly 52 false 0 - - monthly 84 false 0 - -
Protect against possible data corruption
Unified replication limits the contents of the baseline transfer to the Snapshot copy created by SnapMirror at initialization. At each update, SnapMirror creates another Snapshot copy of the source and transfers that Snapshot copy and any new Snapshot copies that have labels matching the labels defined in the Snapshot policy rules.
You can protect against the possibility that an updated Snapshot copy is corrupted by creating a copy of the last transferred Snapshot copy on the destination. This “local copy” is retained regardless of the retention rules on the source, so that even if the Snapshot originally transferred by SnapMirror is no longer available on the source, a copy of it will be available on the destination.
When to use unified data replication
You need to weigh the benefit of maintaining a full mirror against the advantages that unified replication offers in reducing the amount of secondary storage, limiting the number of baseline transfers, and decreasing network traffic.
The key factor in determining the appropriateness of unified replication is the rate of change of the active file system. A traditional mirror might be better suited to a volume holding hourly Snapshot copies of database transaction logs, for example.