This post is more than 5 years old
522 Posts
0
911
VMware SRM with RecoverPoint and XtremIO
Hi,
Just curious what people might be doing out there or what the messaging is around using RecoverPoint with SRM, but having the underlying arrays be XtremIO (specifically the target). Utilizing snapshot shipping, the replica devices are not "WD" or "No Access" as they typically are with the other splitters and this means that leaving them mapped and provisioned to the target ESX clusters for VMware SRM purposes adds a little more complexity or knowledge around knowing not to touch them. Unmapping them outside of SRM testing is likely not going to be an option since it defeats some of the purpose of DR automation with SRM, but wasn't sure what people are seeing as the preferred approach?
thanks!
Idan
675 Posts
0
August 27th, 2015 06:00
Hi Keith,
That's correct, as of RP 4.1.2 and XtremIO 4.0, the replica volumes will remain accessible so caution must be taken if continuously accessing the replica volume (This matter is widely mitigated by the fact that the snaps on the target XIO array are private to the user which RP uses and tech).
If SRM with RP SRA is managing Test and Failover recovery tasks then just make sure to use SRM which will guarantee the DSs are used after image access has been executed. If accessing the target datasores directly then make sure to manually enter image access (after disabling SRM external management) or even better unmount datastores if they aren't needed. You are right in saying that SRM cannot map the replica volumes as they need to be mapped to the ESXs on the replica side.
Additionally, in the near future we are planning to introduce the capability to put the replica volumes in NoAccess device state while not in image access.
Regards,
Idan Kentor
RecoverPoint Corporate Systems Engineering
idan.kentor@emc.com
echolaughmk
522 Posts
0
August 27th, 2015 06:00
Thanks Idan,
For the first part, my understanding is that even though the snaps are private from user intervention, the replica volumes that would normally be presented and left presented to the ESX servers on the DR side are still R/W which could allow an admin that might not be that familiar with the setup to think they can use those as free disks and just start to create datastores and what not on them. I don't think they would be blocked from doing so correct? If the snap promotion happens every 30 minutes on the back-end of the DR array, then its possible that 30 minutes after a junior admin uses the datastore mistakenly, that the snap promotion will overwrite the contents of that....probably causing some issues on the host or anything running on that datastore.
If what I said above is correct, then it still puts all of the onus on the end user not to use those datastores regardless of how the internal RP snapshots are protect by user access.
Let me know if my understanding is wrong....feel free to drop the 411 on me
. That's good news about the enhancement..thanks for that info!
Thanks!
I think SRM will also fail if those replica's are in use by accident ahead of time before the user invokes an SRM test or recovery. I don't think it will just seem them online and promote the snap to overwrite them and move on with the recovery plan from my previous testing, but haven't gone through it with XtremIO yet.