Unsolved
This post is more than 5 years old
5 Posts
0
771
Disaster recovery exercise with multiple vdm failover
Hi,
We are trying to conduct a disaster recover exercise comprising of several VDM ( virtual data movers) and many PROD File systems failover to a secondary site. My question is -- does EMC supports multiple vdm failover at a time or the recommendation is to failover one VDM at a time even thought the VDM's are spread across multiple datamovers? This is imperative to know as the team is trying to reduce the RTO associated with failover and failback comprising with large number of File systems, any replies on it is appreciated.
Thank you!
Rainer_EMC
8.6K Posts
0
October 24th, 2017 06:00
how are you replicating ?
sync vs async is very different in terms of implementation
protocols used? NFS? CIFS?
csubhash
5 Posts
0
October 24th, 2017 11:00
We are using VDM replication consisting NFS and CIFS.
The VDM's has file systems which are also multiprotocol used as NFS and CIFS.
csubhash
5 Posts
0
October 26th, 2017 10:00
Kindly update the string
Rainer_EMC
8.6K Posts
0
October 27th, 2017 02:00
I would suggest to create a test vdm and file systems and test extensively before trying on production.
and see the VNX replication PDF manual and knowledgebase articles
I believe VDM and fs failovers are serialized
plus VDM and fs failovers are separate (unlike Unity which has group failover)
In terms of failover transparency keep in mind that VNX async replication by default uses a different IP for the dst so NFS clients will have to remount
There are ways around this but they require manual work and deeper understanding.
IF all you want is to check the dst without actually failing over you can do that by creating writeable checkpoints on the dst and make them available through a global CIFS server on the dst VDM
If you need more help I would suggest to engage EMC professional service
csubhash
5 Posts
0
October 27th, 2017 11:00
Thanks Rainer for providing the valuable information.
I am well versed with the VDM replication and FOFB statistics what i haven't done in the past switchover / failing multiple VDM's and File systems at a time. I was trying to understand even if these vdm's are spread across multiple Datamovers do we still follow the methodology of switchover / failover of the VDM's and File systems at same time or they have to one after another.
umichklewis
1.2K Posts
0
October 30th, 2017 13:00
As a former PSE, when I produced cadence documents for customers, I recommended failing over each VDM in-sequence, then failing over the associated filesystems. Ergo, failover VDM_1, then FS1, FS2 and FS3, followed by VDM_2 and FS4 an FS5, etc. etc.
There's no reason why you couldn't just failover VDM_1 and VDM_2, then do FS1 through FS5. There's no operational reason why one is better than the other, it's just a matter of choice. Thanks!
csubhash
5 Posts
0
October 30th, 2017 14:00
Thank you Karl, your answer definitely helps us here and preparing for the DR exercise. This is ideal since you were in PS and have experienced used cases.