Hello!<br /><br />We have a server with 3 2.8GHz xeon cpu<img src='/community/emoticons/emotion-7.gif' alt=':s' />, 3GB ram.<br /><br />The server is running a web application and recently we made a change, but the results is not quite what I expected, can someone help me figure this out?<br /><br />We changed one table, so that we made the clustered index (also primary key) on that table only have one column instead of two. The change gave us better results in all of our queries, since all of our queries were just asking on one column instead of two.<br /><br />We run the performance monitor with a frequence of 900 seconds.<br /><br />Before the change our max on CPU were around 70%, and avarage was 42%. <br />After the change the MAX raised to 93% and avarage was about the same (43%).<br /><br />When I run a trace I clearly see the improvement of speed, since the queries that earlier were taking up time is now running very quickly.<br /><br />Before the change we had about 170 transaction per second, and after the change it was raised to 197 TPS.<br /><br />The biggest change is the user connection value. Earlier we had max 691 user connections, but after its 397.<br /><br />We still have the same amount of queries, and the same amount of users connected.<br /><br />I can see that the user connection probably is lower because of that the web pages loads more quickly and does not have to stay connected that long anymore.<br /><br />But why is the MAX CPU higher now? Earlier when the queries took much longer, shouldnt it have been working on the queries then? Or can it be that it was alot of waiting and other things that it had to handle before?<br /><br />The result for the end users is what we expected. Its alot faster, so that is good.<br /><br />This is a good example of that you should not only check the CPU load when trying to fix performance issues.