The first thing to do when you want to improve your SQL performance is you need to learn how to optimize your system by finding out how much hardware you really need to run your applications. The best way to tune your hardware and monitor performance is through the art of performance monitoring which takes experience, knowledge, and sometimes even luck.

Performance monitoring guidelines:

Make sure you’re running your typical processes and work loads during the monitoring.

Don’t only do a real-time monitoring of your servers; capture long running logs.

Always have the disk counters turned ON.

Set up the chart windows with an interval of 18 seconds for routine, daily desktop monitoring.

Know the tools you are working with.

Don’t be afraid to experiment.

Know the terminology (“objects” are lists of individual stats available; “counter” is a single stat; “instance” is further breakdown of a counter stat into duplicate components).

A bottleneck happens when the hardware resources can’t keep up with the demands of the software. This is usually fixed in one of two ways: first, you identify the limiting hardware and increase its potential (i.e. a faster hard drive or increase the speed of the computer); second, make the software processes use the hardware more efficiently.

Five areas to watch when improving SQL performance and identifying bottlenecks:

Memory usage
CPU processor utilization
Disk input/output performance
User connections
Blocking locks

Marisa Pellegrino is a freelance writer from Montreal and is the head researcher and content manager for www.sql-recovery.com.