VNX Release 32 and VDI use case

Hi Bill

Q: does anyone know which VNX OE code version will contain a permanent fix for this?

A: None currently, however there is a hotfix that is available which can be applied over rel 32 P204/Rel 32 P206 - the hotfix is available from EMC Engineering via support

Q: Since VNX arrays are all shipping with Release 32, does EMC want these customers to "downgrade" to Release 31?  Or should VDI users just hold tight and wait for a permanent fix?

A: This is being fixed in future service pack of Rel 32, so really the question is how impacting is the issue for your VDI environments. If it does impact in terms of performance then certainly request the hotfix

Q:How serious is this bug?

A: Basically can cause performance degradation, its an open question so the answer really is depends

Hope this helps answer your questions

Best Regards

Gearoid    

Hi Bill

Ill be honest my exposure was from customers who had performance issues on Rel 32 after the upgraded from Rel 31, so it was more clear that we could see fast cache was not behaving properly

Engineering are aware of the issue literally all you need to say is VDI/FastCache/Rel 32 in the SR and ensure that support goto engineering and you should get the hotfix, it helps to quote the KB article also

Gearoid

Gearoid,

Thanks for the great info.  I have a number of customers on P201 and a few on P204 using VDI workloads (VMware View and Citrix) but none of them have necessarily experienced major performance issues at this point, so it's difficult for me to know if they are affected by this bug.  I guess my overall question is, if this bug manifested itself, would we know for sure?  Secondarily, is it wise for us to proactively apply this hotfix, even if we don't know if we are experiencing this bug?  Finally, in case I need to request this hotfix, what is the name or number of this hotfix for me to ask EMC Engineering in the SR?

Hi to all,

few questions about this topic:

1) Which VDI software is affected? XenDesktop? Vmware VIEW?

2) What about disabling FAST cache for VDI Luns? Can this action allievate this issue?

thanks in advance

In my opinion, this would affect any VDI setup as the problem is heightened due to the random small block I/O nature of the workload and it potentially not being promoted into FAST Cache.  This would technically be the same for XenDesktop or VMware.

Turning it off would make things far worse,  i.e. at the moment you are getting some benefit from FAST Cache,  i.e some blocks are still being promoted into FAST Cache, some are not due to the filter driver deployed as part of FLARE 32.  If you turn it off you will get no FAST CACHE benefit and everything would slow down, i.e. workload would be constrained to spinning disk only.

Hope that helps

Does anyone know if this affects all releases of FLARE 32 or is it only the ones mentioned from a hot-fix perspective?

So for example if i have a customer who is on a release lower than 05.32.000.5.201,  they have the problem as they are on FLARE 32 but there is no hotfix available so they have to wait for the path release (MR1 SP2) I believe.  Can anyone confirm if that is the correct understanding of the situation,  it is not clear from the ETA.

If you are currently using VDI and running Release 31 FLARE code, EMC recommends waiting to upgrade the storage array's code to Release 32 until a service pack is released that contains the fix.  The fix for this issue is not yet available in Release 05.32.000.5.206 (MR1 SP1.1); however, a hot fix is available and can be obtained by contacting EMC Technical Support and referencing this EMC Knowledgebase article.

Hot Fixes are available for:

05.32.000.5.201

05.32.000.5.204

05.32.000.5.206

It affects all Rel 32

The hotfix is only available for the codes

05.32.000.5.201

05.32.000.5.204

05.32.000.5.206

To the best of my knowledge affects all virtual environments (Hyper V for example also)

Gearoid    

This issue is addressed in Release 05.32.000.5.207 and above