Performance issue

The commands are under symqos.

look in the SE_73_ARRAY_CONTROLS_EV.pdf file.

Use the modify argument to change the values of a cache partition, as follows:

symqos -cp -name PartitionName -ran-sid SymmID

modify [-target TargetPercent]

[-min MinimumPercent]

[-max MaximumPercent]

[-wp WritePendingLimit]

Thanks Robert, but I have a question.

What is -ran-sid. can you please explain

If you have Solutions Enabler installed, try running symqos -h to get the syntax and explanations for the symqos parameters. -sid is the parameter for the symmetrix serial number. -ran looks to be a typeo (at least at Se V7.3)

Hi,

This is in response to your first query:

Could you please throw more light on the following information please:

• Is FA reservation is enabled?
• Is the Symm is supporting staging in addition to production data?
• What is the current cache capacity?
• Is the Symm a fully populated array?

These information are rendered helpful in getting an estimate of what is causing the high WP issue on the Symmetrix.

• Also, could you please help me understand what RAID (if uniform across the array) are you deploying in your environment?
• Have you run performance analyzer to pin point the issue to a specific time stamp on every day or so?
• Also, are you configuring Timefinder SNAP\Clone\BCV for any backup related jobs?
• What kind of replication, if any have you configured in your environment?
• Any ongoing migration activity from\to the Symmetrix?

Please throw light on the above so that we can get an understanding on the reason why the WP is high.

Many thanks in advance,

SreeHari

In general, hitting system WP limits is caused by one thing, writing to cache faster than the backend can destage.

You can correct this by slowing the writes

Or by increasing the destage rate by:

1) Adding DAs or disks

2) Using faster disks

3) Changing RAID protection to one with less overhead

DCP can help isolate some volumes that are doing heavy writes so they don't impact all volumes.

At present. the modification is

target % 65, Minimum target 0 %, max target 100 % , donation time (sec) 37  wp limit % 80

I think here we misses is 300-600 time right?  evertime we tweak at least every 3 hours. Can you clarify on this, what is your recommendation. you mean to say  just changing time to 300 600 will work? What do you recommend? with the present parameters, to have a permanent solution?

Yes, What we are doing now is just modifying the DCP, whenever SATA ----%WP peaks. EMC recommends any permanent solution for this issue? We are not planning to expand DMX-4 at this point.

reply to SreeHari

1.FA reservation is enabled

2. no staging

3.current cache capacity 200 GB

4.Yes almost populated. no more disk addition possible

5.Raid1, Raid6 and Raid 5

6.Yes, 8:00 am until 5:30 pm

7.yes. BCV

8.No other  replication only SRDF/s

9.No migration activity

Thanks so much, awaiting your inputs.

Are you already using DCP?

Why modify it?  Are you using the "dynamic" part of DCP?

I would think setting the SATA devices to min = 0 target = 10 max = 20-100 and time something like 300-600 should work all the time without having to tweak it at all.

The donation time won't impact the WP control at all.  I think that target vaule of 65% is way to high.  For controling WP counts for slow drives I usually recommend 10-20% at most.  Also a donation age of 37 seconds is pretty small. 600 seconds is pretty reasonable.  That controls how old a READ slot must be before some other partition can take it.

By the way, in case you don't know, the target value is what sets the WP limit for a partition, not the max or the min.