Start a Conversation

Unsolved

This post is more than 5 years old

1808

August 4th, 2015 20:00

FAST VP VMAX 10K new LUN is always slow

Dear Gurus,

I would like to ask your advices.

Every time we create a new LUN in our VMAX 10K, and we vmotion (VMware) our existing SAP systems to the new LUN, it seems the new LUN needs to take few days to reach the same performance that we had in our old LUN. For the first few days, the performance in the old LUN seems so much slow and all our users complain about it.

Is this logic correct, and if yes, anybody can advise us what to do?

Thank you.

Regards,

Vonk

1.3K Posts

August 5th, 2015 06:00

I can think of a few things that may be impacting the performance with a new LUN vs. an older LUN.

1st is cache.  It may take time for cache to get populated and to get higher read hit rates.

2nd is FAST.  It takes time for FAST to promote extents based on workload.  Where are you binding these new LUNs?

3rd is allocation overhead.  Until all the extents are all written they will have slightly higher response times.

1.3K Posts

August 5th, 2015 19:00

What tier are the new devices bound to?

August 5th, 2015 19:00

Thank you so much Quincy56 for the answers. Our system seems ok now after 2 days. That's a bit slow, but finally it is ok.

122 Posts

August 5th, 2015 23:00

To which disk technology the LUN is bounded, FC, SATA , etc ?

August 5th, 2015 23:00

I am sorry but I don't get you of the tier. It is in the same storage box as the old one if that's what you mean.

122 Posts

August 6th, 2015 00:00

Thats might be the reason and I am sure now the extents wont be in SATA tier

August 6th, 2015 00:00

Oh ok, sorry. it is on sata.

August 6th, 2015 00:00

Sorry, i am quite new on this one, but we actually use FC and EFD with EFD larger than FC.

Thanks.

122 Posts

August 6th, 2015 01:00

There are some consideration prior LUN allocation, please check them before you proceed.
For achieving better performance

August 6th, 2015 02:00

Will do.

Thanks.

1.3K Posts

August 6th, 2015 07:00

There was a feature added to FAST VP called allocation by policy.  Prior to that feature if the bound pool ran out of space, allocations would fail.  With the feature, new allocations can occur in any tier that is in the FAST VP policy.  Because of this, the old recommendation was to bind devices to the pool with the most capacity, usually SATA.  With this newer feature, we do NOT recommend binding to SATA ever.  Bind to the middle tier if you have 3, or to the highest tier, if there are only two tiers in the FAST VP policy.

Binding to SATA will ensure that you get slow performance for initial allocations.

August 10th, 2015 22:00


Thank you Quincy. However I come to realise that we are actually using FC and EFD. Our system is doing fine with the normal speed now.

No Events found!

Top