Periodic snapshots and replication

[francisaugusto]

Hi there,

I am a bit confused about how to create replication tasks.
I've created a task to replicate a dataset recursively, but excluded child datasets:

Dataset A has 3 child datasets: B, C and D. I excluded B.

Now, when choosing the source dataset to replicate, should I choose Dataset A, C and D? Or should I choose just A, as it already will have snapshots containing C,D? Will it contain B?

I am a bit confused about how to combine periodic snapshots and replication tasks - can I just have replication tasks and choose the datasets I want? Will it create incremental snapshots?

Really confusing...

 

[gary_1]

There's a few ways to set this up, but one option is create a single (recursive) periodic snapshot of the root dataset A. That will generate snapshots of A and A's children, B,C,D. Then add a replication task for that periodic snapshot but list child dataset B as excluded.

When the replication runs, you'll see A, C and D replicated but not B. If B has child datasets of its own, I'm not sure if they'll also be excluded, I would assume/hope so, but have not tested.

Another option, would be to exclude B in the periodic snapshot task (rather than replication task) and then the recursive replication of A will still only replicate A,C,D since B never had any snapshots created (snapshots would only exist for A,C,D)

Yet another option would be to do a recursive snapshot task on A (creates snapshots for A,B,C,D). Then create a replication task and tick all the datasets you want to replicate and do not tick B. The downside to this method is if you create any new child or root datasets you may forget to include them in the replication task, and not have a backup when you need it. This also assumes you either did not have recursive snapshots enabled, or if you did, you did not include A (as that would recursively include B).

As to which method you should use, it depends on whether you want newly created datasets to automatically be included in a snapshot/replication or not. I set mine up with a could of periodic tasks (2w and 1y retention periods) and then a single replication task that excluded the child datasets I don't want replicated.

 

[francisaugusto]

Thanks a lot @gary_1.

I realized that when setting up a replication task, I am forced to exclude whichever child dataset that was also excluded from the periodic snapshot task. So I can rest now knowing that no unnecessary snapshot is being created.