Another Testing Scenario - Isolating the Recovery Site

[cbucheli]

Question for the gurus out there:

Setup: We have a simple setup with a single protected and single recovery site. The network configuration is stretched VLANs. All testing to bubble networks works as advertised, and we're quite happy with the solution so far. In addition to the Virtual infrastructure there are quite a few physical systems that make up the environment and the goal is to do a full fail over site test in the coming weeks.

Test Scenario: The proposed solution is to drop the link between the protected-recovery sites, and power up the entire failover datacenter in a bubble (only accessible through VPN links). Once the test is complete, we would then back out any/all changes to the recovery site, in order to have it once again protecting the primary site.

Question: Would it be possible to run the SRM portion of the test in "Test mode" only -- to have it bring up the entire enviroment attached to the live VLANs (which are now isolated from production via dropping the site link), and once complete, reset the recovery site per the normal procedure (i.e. "Test recovery complete, press continue to clean up...")? I don't see why this wouldn't work, unless for some reason the testing procedure requires communication between the sites that the actual recovery process does not.

Thanks for any feedback.

[mal_michael · Accepted solution]

Hi,

Test recovery does not require connection to Protected site as far as I know.

Regarding the networking part of your question, when you edit a recovery plan, on networking screen you have an option to configure test networks. These networks will be used during test instead of bubble networks (default).

SRM will cleanup the virtual environment, but how do you cleanup physical servers?

For how much time are you planning to run the environment at the recovery site? Do you have enough capacity for snapshots on your storage array?

Michael.

[jof]

Hi,

I am looking at a similar situation where we will want to test physical and virtual servers in our DR site.

The plan is to:

Break the link between live and DR sites

Run a Test using SRM (re-IP VMs during this test)

Test all servers (physical and virtual) using client to access applications - VMs brought up on DR VLANs, not in bubble

Finish SRM test (Test Recovery complete) - process already in place to clean up physical servers

Reestablish link between the sites to let SAN replication continue

Are there any issues with this from an SRM point of view.

Thanks

[cbucheli]

I can tell you from the test that we ran (which is fairly identical to the scenario that you're outlining) that everything worked swimmingly with one tiny exception; and that was with the SRA sync status and what the SRA was looking for to allow a Test scenario to be run. In our case, the sync status was automatically set to "Stopped" when the link was severed, which forbade the SRA from running a Test scenario, instead requiring us to run an actual failover to complete the testing. Not the end of the world, as we just need to rebuild the PGs and RPs, but not nearly as clean as a "Test" would have been.

Thanks for the feedback. It's good to have a sanity check from time to time.

[jof]

Thanks for the info. It's good to know what to expect.It's not ideal but like you say not the end of the world either just a bit more work.

Did this mean that you needed to do a full resync of the data after the test?