VNX and vSphere update – important VAAI XCOPY changes

For all VNX gen 1 customers (VNX 5100, 5300, 5500, 5700, 7500 running Elias and Inyo code bases) I’m very happy to note that we’ve issued a large update that back-ports a set of fixes that are already in the next-generation VNX hardware (with the Rockies codebase) that I referred to here.

This is in VNX gen 1 Operating Environment For File Version 7.1.74.5 and Block Version 05.32.000.5.209, known internally as “Inyo MR1 SP3”.   It is a “rollup” release that includes Hotfixes that have been in circulation for some time (we use this MR or maintenance release model as mainstream updates).

Specifically for the VMware aficionados out there, this changes the internal handling of operations that touch XCOPY.

We had a regression that affected XCOPY operations sometimes creating a high host-side latency for some customers.  

This was covered in EMC Technical Advisory (ETA) 172796 and KB article 172796.   I want everyone to be clear that it does not affect VNX next-gen platforms running Rockies (which completely avoids internal MLU queues which are at the root of the issue).

There are, of course a ton of additional updates (maintenance releases are like .1 or .5 releases in Microsoft of VMware land).   If you are using VDMs – this can be a non-disruptive upgrade on block and NAS.  If you aren’t using VDMs, there will be a very small maintenance window on the NAS side.   BTW – if you aren’t using VDMs yet – why not?   They are free, do everything the physical DMs do, have no material performance impact, and make NAS upgrades (including across platforms!) going forward non-disruptive.   I’d love to hear from customers that are using them, and perhaps from those who aren’t – why not?

You can get the update from http://support.emc.com

The VMware HCL shows .32.* code as OK but customers should move forward with confidence and consider/plan with EMC or their EMC partner updating platforms to this MR1 release.

Comments

[Amir]

There needs to be a better EMC effort in marketing the VDM (-move) feature , to (finally) say we can do cluster like shifting of resources between active blades. this off course requires at least 2 active blades, and also possibly raises the consideration of dropping the standby passive blade and making it active.
The only downside to that is HW redundancy lost.

[Ian]

Unisphere on gen-1 VNX systems does not allow configuration of NFS shares when using a VDM. We have a lot of Linux hosts, so NFS is critical. For the one array that we have using a VDM, we have to use the CLI to manage NFS shares.