July 20, 2007 at 9:06 am
Hi,
Good day to all of you
Recently I have battling with performance slowness issue happened in my production,which until now I still can't seemed to find the actual root cause of it.
Hope you experts out there could shed some light so I can end my misery once and for all.
Here's the scenario:
We have 3 SQL Servers (SQL Enterprise 2k sp4) running in the production, all hook up to same EMC SAN box with different set of LUNs.
When we perform full database restore with REPLACE option using Litespeed in server A we encountered WriteLogs waits queueing up in server B and C causing the production servers B and C to run extremely slow.
The restore process get the backup file from the X drive within the same server A and restore to G drive in that same server.
We managed to contain the issue by changing the G drive from ATA disk format to Fiber disk.After running for a few months, the issue seems to haunt us again.
Would really appreciate it if someone could tell me what i should do next to resolve it.
Thanks in advance!
July 20, 2007 at 10:11 am
Are you sure that your LUNs aren't sharing physical drives? I'd definitely have the SAN people dig into the configuration and while you are doing a restore, check and see if your I/O to and from disk changes on the other servers.
July 20, 2007 at 10:17 am
Just to expand on what Steve has said, separate LUNs do not always mean separate physical disks. Many SAN administrators don't understand that a database server requires separation at the physical disk level so they just carve you off a piece of available disk space and you end up sharing I/O resources with a file server or exchange server. Your database needs to have its own physical disks/spindles, not just logical separate LUNs.
July 21, 2007 at 12:35 am
I see, thanks guys for the input... is there any tools that can help me to verify that?
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply