SQL 2005 wait event statistics are a good step forward, but are only useful for isolated testing, and to get a global impression on how the SQL Server is spending its time. Mario Broodbakker demonstrates why it would be much better if SQL Server collected those statistics per session and per subtask.… Read more
Measuring what is actually happening is always the best course of action when investigating performance issues on databases, rather than relying on cache hit ratios, or best practices, or worst of all, guesswork. This article introduces some techniques that will allow you to pinpoint exactly where the performance issues are in your system, so you'll know exactly where to spend your time (and money) in solving them.
… Read more
When faced with two viable solutions to a badly compromised database design, one using clustered indexes and the other compound primary keys, Grant Fritchey took the only sensible route: he gathered hard performance data...… Read more