SQL Server - Sql Server 200x Stress Tools?

Asked By sloan

15-Jul-08 10:17 AM

I currently have Sql Server 2005 Standard Edition.

I have a few legacy Sql Server 2000 Standard Edition databases as well.

..

I have implemented the Sql Server 2005 Performance Dashboard Reports as
well.  Those have helped some.

..

I am looking for suggestions on tools which can artificially stress a
database, and capture results, you know the drill.


I have coded up a mini application which calls my BAL DotNet assemblies to
stress the overall system as well.  Basically, it creates threads and sends
in lots of work requests.


Money isn't the issue.  Something that works and works well is.

..

PS
I've been trying to push to get 2005 Enterprise Edition as well, so if there
is something in Enterprise which is not in Standards, please point that out.

Thanks.

15-Jul-08 10:43 AM

Light-weight and free: ostress from Microsoft.  Google it.

15-Jul-08 12:38 PM

As for stress testing the database server specifically, SQL Profiler has a
feature to replay trace logs. For example, the trace can be captured on a
production server and replayed on a development server.
http://blogs.msdn.com/sqlprogrammability/archive/2008/06/26/use-sql-server-replay-tools-to-reproduce-and-resolve-customer-issues.aspx

15-Jul-08 08:19 PM

What about the Database Hammer tool from the SQL Server 2000 Resource
Kit.
Is that still useful?


Denis The SQL Menace
http://www.lessthandot.com/
http://sqlservercode.blogspot.com
http://sqlblog.com/blogs/denis_gobo/default.aspx




On Jul 15, 12:38=A0pm, Eric Russell
a
com/sqlprogrammability/archive/2008/06/26/use-sql-s...
to
sends
there
out.

15-Jul-08 12:50 PM

The SQL Load Test tool can generate unit tests for the Visual Studio Load
Test using a Profiler trace:

http://www.codeplex.com/SQLLoadTest

HTH,

Plamen Ratchev
http://www.SQLStudio.com

15-Jul-08 01:44 PM

Analysis Services works just as well as Database Hammer, if the client is
refreshing their cubes from the production OLTP server each morning from 7am
- 10am.

15-Jul-08 02:37 PM

I don't :-)

But kidding aside, for this to be a useful exercise, you need to first
identify the objective of your stress test. If you just want to know the
general performance characteristics of your hardware, you design your stress
test in one way. If you want the stress to shed a lot of light on how your
apps/databases would perform underload, you design your stress test in a
totally different way. If you just grab some stress tool and hammer the
server away, the results may not be useful at all.

Linchi