Setup & User Guide
97 |
DOC-246-GDE-Pivot3 Acuity 2.3 Setup & User Guide-v1.0.docx
Section 13 Replication
Asynchronous Replication Overview
Asynchronous replication creates a copy of a source volume on a target Acuity vPG. When a snapshot is taken of a source
volume, the source data is copied to an inbound replication volume on the target vPG. Only data that has changed from
the previous snapshot is sent, so replication is faster and impact on system performance is minimized.
Asynchronous replication allows multiple source volumes on a vPG to be replicated in parallel, up to four source jobs per
accelerated node for a total of eight jobs per vPG. When multiple simultaneous jobs are started, they run in parallel on the
two accelerated nodes of the source vPG. Replication will automatically select the pool with the most available space to
host the inbound replication target. If there are multiple jobs with the same performance policy, they run in the order that
the snapshots were taken. The snapshots inherit a performance policy one less than the source volumes being copied.
It’s important to remember that while there is a limit of eight asynchronous replication jobs that run in parallel on a
source vPG, a target vPG may be the target for any number of asynchronous replication jobs – and the target system is
limited to, at most, 20 simultaneous replications. Schedule asynchronous replication jobs carefully to avoid reaching this
limit or saturating the links. Throughput and time spent replicating are optimized by sequencing a few jobs at a time and
not scheduling them to kick off all at the same time.
To verify that the scheduling of replication jobs is optimized to produce minimal system impact, monitor the Long
Running Tasks list. If needed, immediate corrections can be made by canceling tasks. If replication tasks are continuously
being superseded, this is an indication that the replication interval is set too low and should be increased. The longer-
term solution is to change the replication schedules to minimize accumulated jobs.
Snapshots can be retained or deleted based on a schedule or policy. If the retention is set to 10, replicated snapshots will
accumulate until there are 10 successful replications on the target; when the 11
th
replica is completed, the oldest is
deleted from both the source and the target vPGs.
The retention count for a protection policy or schedule only applies to that protection policy or schedule. For example, if a
volume has retention set as follows:
• Daily schedule with retention count of 7
• Weekly schedule with retention count of 4
• Monthly schedule with retention count of 12
The retention for each unique schedule or protection policy is tracked separately, so that for the above volume the
volume would have 7 + 4 + 12 = 23 replication snapshots.
NOTE:
Retention on a schedule does not apply to manual snapshots. If a snapshot is taken manually, it must be deleted
manually. Manual snapshots are not included in the retention count.
The size of the snapshot on the source may be different from the size on the target. This is normal and expected behavior.
There are a variety of factors that can cause this to happen: partially filled write blocks, data reduction on the target,
snapshot frequency on the source being different from replication frequency to the target, for example. More information
on this issue is provided in Acuity Data Usage Reporting.
Replication jobs are prioritized by the QoS level of the source volume. If multiple replication jobs are queued, Mission-
Critical jobs will be run before Business-Critical jobs, which are run before Non-Critical jobs. However, if lower priority jobs
are skipped, their priority is gradually boosted so that they will eventually run and not be queued indefinitely.
To Next Page
Loading...