Re: We quadrupled hardware power and reduced performance
From: Jackie Brophy (jbrophy_at_mathworks.com)
Date: 04/13/04
- Next message: Tom Moreau: "Re: Check constraint"
- Previous message: Dr. StrangeLove: "Re: SQL query to return multiple rows into single row"
- Messages sorted by: [ date ] [ thread ]
Date: Tue, 13 Apr 2004 09:01:54 -0400
I had very similar problems, did update stats etc. The one thing that helped
was to turn parallelism off on the server. Right click on the server,
properties, processor tab, use only 1 processor. This helped in our case
pretty dramatically. As in all cases, test, test, test.
Jackie
"Baisong Wei[MSFT]" <v-baiwei@online.microsoft.com> wrote in message
news:UAyUFACDEHA.2304@cpmsftngxa06.phx.gbl...
> Hi Randolph,
>
> As for all the components work as a whole, we could not be sure that one
> part's improvement could bring a overall improvement. Though performance
is
> always a big issue of SQL Server, we do have some general rules to
> troubleshoot it. Here I just want to add some information on that.
>
> When we face a performance issue, we should have a baseline of it ( for
> your case it would be a hardware improvement should bring a overall
> improvement of the whole system) and we should narrow down the problem to
> the bottleneck from Application/Query/Database Design -> Operating
> Environment-> Hardware. In your case, the later 2 changed, and the whole
> system does not meet you expectation. So, we should make the expectation
> clear: Query slow? CPU high? In you case, it seems that the high CPU
> utilization is your concern, and I am not sure if this has bring a impact
> to your application performace. Then, some tools are needed to check why
it
> is so high. I would say that in some sense, the CPU is driven by the other
> subsystems. In other words, bottlenecks in memory or I/O will often first
> appear as a CPU bottleneck. Because of this ver issue, the best process
for
> evaluation system perfomance is to first analyze memory, thant the I/O
> subsystem and finally the processor. This is an important concept in that
a
> shortage of memory will tend to drive a significant percentage of kernel
> time. This is not to say that you should completely ignore the processor
> when first evaluation system performance. This is especially true in a
> "hang" situation.
>
> There is some counters that you might use to monitor:
> %Processor time
> %Priviliged time
> %User time
> %DPC time
> %interrupt time
> DPCs Queued/sec
> Interrupts/sec
> Processor qurue length, ect.
>
> and then, you'd better determine SQL Server's contribution to the CPU
> bottleneck.
>
> As for more detailed information and method to troubleshooting performance
> issue, I would recommend the following articles for your reference.
>
> HOW TO: Troubleshoot Application Performance Issues
> http://support.microsoft.com/default.aspx?scid=kb;en-us;298475
>
> Support WebCast: How to Collect and Analyze Performance Data in Microsoft
> SQL Server
> http://support.microsoft.com/?id=324692
>
> How To Gather Information for Effective Troubleshooting of Performance
> Issues
>
http://support.microsoft.com/default.aspx?scid=http://support.microsoft.com:
> 80/support/kb/articles/Q175/6/58.ASP&NoWebContent=1
>
> Hope this helps.
>
> Sincerely Yours
>
> Baisong Wei
> Microsoft Online Support
> ----------------------------------------------------
> Get Secure! - www.microsoft.com/security
> This posting is provided "as is" with no warranties and confers no rights.
> Please reply to newsgroups only. Thanks.
>
>
- Next message: Tom Moreau: "Re: Check constraint"
- Previous message: Dr. StrangeLove: "Re: SQL query to return multiple rows into single row"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
|
