jacobymatt
asked on
Proving performance increases on SQL Server
I am the new database administrator for a company running SQL 2005 Standard Edition. The first 2 things I have done were moving the database files to separated drive than the operating system, which should have increased performance a little, and then I changed the nightly automated task that used the old dbcc command to rebuild all the indexes to the new rebuild index task that is inside the maintenance plan for sql 2005. I thought from the combination of those 2 changes I would see some decent increases to performance, but I think my problem is knowing exactly where to look so that I can prove the benefits of what I have done to my supervisor.
I had previously ran traces against the database in question using SQL Profiler and saved a trace before I moved the db files, after I moved them but before changing the index rebuild method, and the once more today with all my changes. So my thinking was I would have 3 different traces to compare, and the last one should look the best as it has the changes made to the database that should have improved performance. I've been mainly looking at the duration column of the statements that seem to be taking the longest. I was hoping to see the duration period decrease and that would be proof of performance increase, but I'm not really seeing that, so I don't have much to show my supervisor right now.
Can someone point me in the right direction? I need to have something beneficial to show today!
I had previously ran traces against the database in question using SQL Profiler and saved a trace before I moved the db files, after I moved them but before changing the index rebuild method, and the once more today with all my changes. So my thinking was I would have 3 different traces to compare, and the last one should look the best as it has the changes made to the database that should have improved performance. I've been mainly looking at the duration column of the statements that seem to be taking the longest. I was hoping to see the duration period decrease and that would be proof of performance increase, but I'm not really seeing that, so I don't have much to show my supervisor right now.
Can someone point me in the right direction? I need to have something beneficial to show today!
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Hi,
The execution duration for individual statements is dubious at best. That is, given the multi-user nature of production SQL Servers, unless that it is a dramatic difference, it could be that the conditions on the server were different between the two runs - some other use was running a particularly complex query with a large result set during your first test run vs the second one.
To answer your question, I believe that the duration column is milliseconds.
Regards
David
The execution duration for individual statements is dubious at best. That is, given the multi-user nature of production SQL Servers, unless that it is a dramatic difference, it could be that the conditions on the server were different between the two runs - some other use was running a particularly complex query with a large result set during your first test run vs the second one.
To answer your question, I believe that the duration column is milliseconds.
Regards
David
ASKER
so to follow up on the document that you suggested reading above. I went ahead and printed that and will be studying up but yes I am the administrator for a 2005 SQL db so I just want to make sure that this will still be worth reading since it is intended for 2000 administrators.
ASKER
Thanks!!!