Start a Conversation

Unsolved

This post is more than 5 years old

2091

April 26th, 2012 12:00

IO Performance issue on LUN assigned to SQL Server (Error: I/O requests taking longer than 15 seconds to complete)

Hi,

DB team is reporting below error on SQL Server (Which is a Virtual Machine) LUN (Tier3 regular device not a thin device). This Lun is provisioned from VMAX and allocated as RDM to the SQL Server (VM).

Error :

SQL Server has encountered 2 occurrence(s) of I/O requests taking longer than 15 seconds to complete on file in database [BESMgmt] (6).  The OS file handle is 0x0000000000000A9C.  The offset of the latest long I/O is: 0x000000006ac800,

I have fetched "Sampled read/write response time (ms)" graphs from ECC and found that read response time for the specific symdev has crossed 80ms at some times in the Graphs.

Can anyone please help us in resolving this issue ?

1.3K Posts

April 26th, 2012 12:00

Some of the ECC utilization numbers can be off, as far as I know you can get SPA for no charge if you are an ECC customer. 

April 26th, 2012 12:00

SPA is not installed.

But we have ECC performance manager in our environment.

1.3K Posts

April 26th, 2012 12:00

Also do you have SPA installed?

If so, I would start by looking at component utilization, FA,DA and disks.

1.3K Posts

April 26th, 2012 12:00

There are about 1,500 Symmetrix performance gurus out in the feild.  You should start with your local account team to see who can assist. 

April 26th, 2012 13:00

ok. I will talk to my team and get SPA installed. But for now we need to get this IO Response issue resolved. Can you please help me getting this IO response issue resolved ?

1.3K Posts

April 26th, 2012 14:00

Can you start by posting some of the TTP/BTP files from ECC? 

April 26th, 2012 15:00

Please provide me FTP link where i can copy TTP/BTP files.

1.3K Posts

April 26th, 2012 19:00

April 27th, 2012 09:00

TTP.zip is copied to FTP.

1.3K Posts

April 27th, 2012 12:00

From a quick analysis, I would start with spreading the load from 8E, 9E and 10E over more of the unused FAs.

Also there are some drives that are very busy.

I've attached a Symmmerge heat map.  The red boxes around the FAs show high queuing.  And the drives are pretty obvious.

UtilizationMap.jpg

May 1st, 2012 22:00

Thanks for informing about load on FA Ports.

SQL Server which is facing slow performance issue is connected to FA-8E:0 . Many other servers are also connected to same FA port. But only SQL server is reporting the IO Performance issue. 

Also can you please advice from where we can generate a HeatMap simillar to your graph as currently we don't have SPA installed for VMAX. 

No Events found!

Top