ONTAP Discussions

SVMDR Test environment

lcoppitelli
2,781 Views

Hi folks,

More than a few times i had a customer ask me to bring online SVMDRs in test environments and separate network.

Problems arises when they also would like to keep the replication going.

This was much easier in 7-mode having to just flexclone the volumes and recreate the vfiler from the vfiler_root and the replication would happily continue on the parent volumes.

As i understand it, right now we are only able to either:

  • Initialize a new replica on a secondary "bin" SVM, effectively doubling the space needed to keep duplicate data.
  • Flexclone the volumes from the svmdr to another svm, saving on data usage but having considarably more effort to correctly re-setup the svm's policies\shares\namespace.

Also you cant create an svmdr relationship from a destination svmdr (on the same cluster) and you cant create more than one svmdr relationship from a source.

 

I remember reading about an SVMDR "test" feature, but i can't find any reference of it anywere,

Am I missing something?Cat Frustrated

2 REPLIES 2

Damien_Queen
2,766 Views

I suppose you can either:

1) brake replication relationship occasionally (which is not the best idea), and bring your volumes online in this way, and then reestablish relationships & continue replica. But in this way along the side from data testing, you'll test your DR behavior like in the real world.

2) or make SnapVault to third storage or the same DR storage and make FlexClone from each (or required) volumes, now you can test them either with the same subnet or different. In this way, you will have basically 3 data sets: one on primary, one in SVM-DR and one in SnapVoult relations. Clones don't need to be split.

lcoppitelli
2,735 Views

Damien,

Thank you for your input but I need to keep an RPO: i cannot keep the relationship in a broken status untill all the application tests are completed, so option 1 is out of the equation.

As for option 2, i already explored that possibility in my post, but doing that would mean having to host duplicate data, which is not always doable.

I'm talking specifically of the SVMDR functionality which keeps all other configuration, not just the raw volumes.

Public