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.
| Author |
Topic |
|
Kristen
Test
22859 Posts |
 Posted - 2010-06-18 : 10:43:47
|
| We are getting close to rollout for a new project.Dedicated box (i.e. just the one database), loads of CPUs, loads of RAM, loads of Disks ... but the box feels "sluggish" to me.Even when I'm the only person on it, and the CPUs are at 0% and the RAM in use is "modest" there is a definite "wait" before each action.I want to run some sort of test to see how it performs. I was thinking of some sort of transactions-per-second or somesuch where I can compare my box to some published results and get some comfort that throughput is reasonable by comparing against comparable boxes.Any suggestions? |
|
|
robvolk
Most Valuable Yak
15732 Posts |
Posted - 2010-06-18 : 10:58:56
|
| http://support.microsoft.com/kb/944837Ostress is the (semi) supported MS utility for stress testing hardware. It's not hard to use but it's not intuitive at first.You'll also need to read up on SQLDiag, which has been included with SQL Server since 2005. BOL has info but I'd recommend reading up on it here:http://blogs.msdn.com/b/psssql/You should also research READTRACE and SQLNexus. They are included with the RML Utility download. The documentation is OK but not the greatest. SQLServerPedia has some stuff on it too.If you need to test the disks, use these:http://www.microsoft.com/downloads/details.aspx?familyid=9a8b005b-84e4-4f24-8d65-cb53442d9e19&displaylang=enhttp://support.microsoft.com/kb/231619More info can be found here:http://sqlserverpedia.com/wiki/SAN_Performance_Tuning_with_SQLIOhttp://sqlblog.com/blogs/kevin_kline/archive/2007/06/28/understanding-sqliosim-output.aspxhttp://blogs.msdn.com/b/sqlserverstorageengine/archive/2006/10/06/sqliosim-available-for-download.aspx |
 |
|
|
Kristen
Test
22859 Posts |
Posted - 2010-06-27 : 17:48:11
|
| Sorry Rob, for some reason I didn't see your reply originally.I'll have a look at this in the cold light of day! tomorrow |
|
|
|
Kbalz
Yak Posting Veteran
94 Posts |
Posted - 2010-06-29 : 11:32:38
|
| Any virus protection might be scanning sql folders? |
|
|
|
Michael Valentine Jones
Yak DBA Kernel (pronounced Colonel)
7020 Posts |
Posted - 2010-06-29 : 12:34:05
|
| I test every server, except virtual machines, with SQLIO. There seems to be a lot of variation in the performance of disk IO and I have had more problems with this than any other hardware cause. I don’t test virtual machines because they are on SAN and the testing can have a serious impact on all the other VMs.I have a standard test script that I run, and I leave it installed on the server so that I can test it again if I suspect IO problems later.I use a fairly large test file size for SQLIO, at least 8 GB, to make sure that cache on the controller is not making the disk look faster than it really is.It is a bit tricky testing on a SAN because of the impact of SAN cache and other servers that may be sharing the same disks or switches. I try to run the tests at off hours, but I never really know what else is going on with the SAN and testing can impact other production servers.CODO ERGO SUM |
|
|
|
|
|
|
|
|
How to Speed Test a Server?