We've been having some minor performance problems with a client's MCMS website with a template that pulls content from a SQL database.  The site would occasionally report SQL time-outs in the data-layer.  I finally had a chance to do some more detailed testing today and have found what I believe to be the culprit, or at least a clue to the root cause.

To force a load on the website, I used the Microsoft Application Center Test.  I watched the CPU utilization on the IIS and SQL servers, no problem there.  Network traffic between the two was minimal.

Figuring that the bottleneck is ultimately somewhere in the SQL server, I started to play around with the Performance utility in the Administrative Tools.  At first I monitored several SQL objects to no avail.  Then I thought I'd look at the disk performance itself.  After experimenting with a few, I found 'Physical Disk: Avg. Disk Queue Length ' was spiking and maxing out at 100 most of the time.  Searching the web, I found http://www.sql-server-performance.com/sql_server_performance_audit2.asp which suggests that the meter should be down around two.

I've since called up the ISP that hosts our servers, and I'm waiting to hear back as to the cause of this excessive load.