Please start any new threads on our new site at https://forums.sqlteam.com. We've got lots of great SQL Server experts to answer whatever question you can come up with.

 All Forums
 SQL Server 2005 Forums
 SQL Server Administration (2005)
 wait_duration_ms & Performance.

Author  Topic 

Hariarul
Posting Yak Master

160 Posts

Posted - 2007-12-06 : 04:20:27
Hi all,

We have 2 SQL servers Development & Production servers with below configurations.

System Info:

Database Version - SQL Server 2005
Database Size - 120 plus GB
OS Platform - Windows 2003
Database Load - 50 users
CPUs - 4
RAM - 8 GB
Bit - 64 bits

We are having a big performance difference between these two SQL Servers.

The Development server is working fine and has a good response time. The Production server ( has Itanium processor) is showing a very very bad response time.

Example:

A stored procedure that takes just 4 seconds in the Development server is taking 8 minutes , 4 minutes , 17 minutes , 2 minutes randomly in the Production server.

The work load , indexes remains the same. The statistics are updated. The recovery model is same. The SQL server agent is also disabled in the production server. No replication in the production server.

Findings :

Finally , When we checked the DMV 'sys.dm_os_waiting_tasks' of the two servers , we were able to see a very huge difference in the column 'wait_duration_ms'.

The difference is 739776235 milliseconds. That is the Production server's 'wait_duration_ms' value is 739776235 milliseconds higher than that of the Development server.

Note : The Production server was a newly built and was up and running just before a month. The Development server is running for more than 6 months.

Any thoughts based upon the above findings ?

Any help would be appreciated.

Thanks & Regards,

Hariarul

RickD
Slow But Sure Yak Herding Master

3608 Posts

Posted - 2007-12-06 : 06:37:20
Yes, have you checked the network cards? You may want to start looking to see if you have an issue there.
Go to Top of Page

Hariarul
Posting Yak Master

160 Posts

Posted - 2007-12-06 : 06:42:20
Thanks Rick. Do you mean this issue would be straight related to the network cards ?
Go to Top of Page

RickD
Slow But Sure Yak Herding Master

3608 Posts

Posted - 2007-12-06 : 06:45:41
Quite possibly. If you haven't opened the ports correctly or set the card up properly, then all data requests will queue until they can be returned.

There are also other things to check, such as the disk I/O, the % processor time, disk queue length etc, but I would start with checking the server (including network cards, firewall and genberal networking) is set up correctly.
Go to Top of Page

RickD
Slow But Sure Yak Herding Master

3608 Posts

Posted - 2007-12-06 : 06:48:11
Also check you haven't limited something in sp_configure as I am guessing you switched on the AWE (You must have if you have 8GB ram).
Go to Top of Page

Hariarul
Posting Yak Master

160 Posts

Posted - 2007-12-06 : 07:00:21
Thanks Rick. We will check the parameters you have mentioned. BTW , we have not switched on AWE for both the machines. Should we do it ?
Go to Top of Page

Rovastar
Starting Member

38 Posts

Posted - 2007-12-06 : 07:00:26
How are other queries?

Select * from mylargetable takes x time on dev is it 10*x (or whatever)on prod?

I am confused is your production sever 'live' if so. How did it fair in the testing/staging before going live? in stress testing, etc
Did this sp perform ok then?

Have you tested this on no load?

Has the sp in production always taken ages or is this new thing?

What does the execution plan imply the problem points are? It may point to a part in teh process that is slow. Mayeb a table is not indexed properly or something.

If it is throughout - all different queries are similiarly problematic then you can rule out specfics that sp, etc.
Go to Top of Page

SwePeso
Patron Saint of Lost Yaks

30421 Posts

Posted - 2007-12-06 : 07:06:16
quote:
Originally posted by Hariarul

The difference is 739776235 milliseconds.
This is 205h 29m 36s (8 days, 13 hours, 29 minutes and 36 seconds).



E 12°55'05.25"
N 56°04'39.16"
Go to Top of Page

Hariarul
Posting Yak Master

160 Posts

Posted - 2007-12-06 : 07:22:59
Thanks Rovastar & Peso.

Rovaster, your questions really makes sense.

It is not always 10*x time duration in the Production server comparing to the Development server. It differs and varies randomly.
As I have mentioned earlier , a same procedure takes different durations when executed consequently in the Production server( 9 mins, 2 mins, 7 mins, 32 mins :-( ) . But when the stored procedure is executed for the same load in the Development server , it runs for less than 5 seconds.

So, The issue is not only with a single stored procedure. Almost all the procedures in this server is taking time.

We are right now in the UAT phase and not gone LIVE.

Main point to note :

The testing , stress testing , performance testing etc were done on the same Production server. But the Production server was re built from the scratch ( OS , SQL Server etc ) before the UAT. So , we could say that this issue suddenly poped out on the same machine after the re built..

Peso,

Does the 205 hours difference gives us any clue for the performance issue ?

Thanks,

Hariarul
Go to Top of Page

SwePeso
Patron Saint of Lost Yaks

30421 Posts

Posted - 2007-12-06 : 07:32:57
Not more than it seems very high.
You had a query that run for more than 8.5 days and no one noticed?



E 12°55'05.25"
N 56°04'39.16"
Go to Top of Page

pootle_flump

1064 Posts

Posted - 2007-12-06 : 09:51:56
What were the spids of the long running tasks?
Go to Top of Page
   

- Advertisement -