The best way to optimize performance of a cursor is, of course, to rip it out and replace it with set-based logic. But there is still a small category of problems where a cursor will outperform a set-based solution. The introduction of ranking functions in SQL Server 2005 has taken a large chunk out of that category – but some remain. For those ...

Greg Linwood, a fellow SQL Server MVP, has started a series of articles in which he attempts to prove that having a clustered index on each table is not a good practice. However, he has failed to include the effects of fragmentation into account, so I decided to run some tests for myself. One of those test had rather upsetting ...