HPE Storage Users Group
https://3parug.com/

Question about a desaster recovery simulation
https://3parug.com/viewtopic.php?f=18&t=2630		 Page 1 of 1
Author:  Spocky [ Mon Aug 28, 2017 2:49 am ]
Post subject:  Question about a desaster recovery simulation
Hello everybody,
I have a question which needs a long explanation ;-)
my company plans to buildup server and stoarge capacities in a second datacenter for disaster recovery.
Between both datacenters there is a FCIP Connectiom
In both datacenters there is also a 3PAR 7200 . We use remotecopy to replicate some mission-critical vms to the Backup location.

Twice a year we do a firedrilltest to simulate a disaster ( first one was last sunday )

after the simulated breakdown I promoted the RC Groups with
„setrcopygroup  recover  -pat RC_GROUP*“

At the end of the test I have to discard all changes done at the backup 3par within the RC LUNs and demote the RC Groups ( ideally befor both array can see each other again )

Now my question :
Is there better approach than deleting and rebuilding the whole RemoteCopy Config ?

Regards
Stephan
Author:  aberus [ Mon Aug 28, 2017 12:26 pm ]
Post subject:  Re: Question about a desaster recovery simulation
There is, you can just simply use snapshots on the remote array exporting them to your DR hosts. Using this method you don't have to stop your RC groups and tear down is simple.

1. Perform a manual sync to ensure that the latest data is being used.
2. Take a snapshot of the VV's
3. Export the VV's to the host for testing
4. After testing, unexport the LUN's
5. Delete the snapshots

If you are using Oracle make sure to place the DB into hotbackup mode prior to the sync and take it out afterwards. I have a post on how to script this portion in the scripting section.
Author:  Spocky [ Tue Aug 29, 2017 9:30 am ]
Post subject:  Re: Question about a desaster recovery simulation
In the case of a desaster recovery simulation the connection will be down and also the remotecopy groups .
So snapshots will be created autmatcally as a I recall.

But your procedure should also work with stopped remotecopylinks ?

We talk about VMware esx and vmfs Datastores
Author:  aberus [ Tue Aug 29, 2017 9:56 am ]
Post subject:  Re: Question about a desaster recovery simulation
You will need to manually create the snapshots but that can be very easily scripted. This is how I've deployed it on my environment and it works great.

You do not need to have the RC links down for this to work nor do you need to stop your remote copy groups, that would only be required if you are bringing the DR systems online with the same IP address as your production servers or you do not have an isolated failover environment. In the case of using the same IP I would "assume" you are using OTV to extend your IP network, but even with that you can get around that by creating an isolated vlan and set your mask to 255.0.0.0 then all of your systems can talk, no matter their IP.

We do not use OTV instead have moved all systems to use DHCP with static reservations, for testing I have an isolated vlan configured that I fail over to. In my case we are a UCS shop so I just associate all of the NIC's on the blades server profiles to use that vlan, in the event of a prod failover I would just change those service profiles. For Vmware I use Zerto which I have configured to fail over to my isolated vlan for testing.

Does this make sense?
Author:  Richard Siemers [ Sat Sep 09, 2017 9:39 pm ]
Post subject:  Re: Question about a desaster recovery simulation
aberus wrote:
There is, you can just simply use snapshots on the remote array exporting them to your DR hosts. Using this method you don't have to stop your RC groups and tear down is simple.

1. Perform a manual sync to ensure that the latest data is being used.
2. Take a snapshot of the VV's
3. Export the VV's to the host for testing
4. After testing, unexport the LUN's
5. Delete the snapshots

If you are using Oracle make sure to place the DB into hotbackup mode prior to the sync and take it out afterwards. I have a post on how to script this portion in the scripting section.


Solid plan Aberus, thank you for sharing. In Step 3 - this implies export of the new snapshots (VVs) in case any readers didn't pick up on that. :)
Author:  aberus [ Mon Sep 11, 2017 6:03 am ]
Post subject:  Re: Question about a desaster recovery simulation
Thank you for catching my omission and pointing that out Richard! It's somewhat of an important detail :)
Page 1 of 1 All times are UTC - 5 hours
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group
http://www.phpbb.com/