Safe to Resync?

[rjkunde]

We are running SW 6.0.4837 w/ the high availability package. Due to a backplane failure in our secondary SAN, it's been out of sync for approx. 2 weeks. When the backplane was failing, we saw weird issues during the sync process (like our production VM's slowing down to a near halt).

Given that the production server is (well... in production ), do you think it is safe to resync the secondary SAN without impacting the end-users?

Were running a 10GBE dedicated NIC between SAN's and 4x 1GBE iSCSI round robin connections also. The disks are 7200 SATA drives, RAID 5 on an LSI RAID card with 16 port expander backplane. I've attached a screen shot of our throughput speeds, but just want to be absolutely sure the sync won't bring our failover cluster to a crawl.

Thanks in advance for your time.

[Max (staff)]

Hi Rjkunde,
I have to say that with an SATA RAID 5 I'm 99% sure that the cluster will get really slow.
According to our current best practices all RAID 5, 6, 50, and 60 arrays are not recommended for use in a 2 node HA environment.
This is caused by the fact none of the above mentioned RAID arrays can sustain continuous IO blender produced by concurrent access of a client hypervisor cluster and sequential read requests produced by the second SAN synchronization process.

So there are 2 options:
1. Fire up the SAN at night so that it doesn't interfere with the cluster (VM shutdown recommended)
2. Rebuild the failed SAN's RAID to a RAID 10 and then turn on the cluster and reconfigure StarWind to use the new array. When everything is resynchronized, rebuild the RAID on the second SAN and repeat the SAN reconfiguration. This will allow you to avoid further storage IO bottlenecks associated with RAID 5/6/50/60.

3. If you're hunting for capacity then deploying a 3rd SAN node may be a good idea. In a 3 node SAN scenario you can use RAID 0 without any risks due to triple data mirroring.

[rjkunde]

Great information. Thanks

[Max (staff)]

You're welcome!