Paul White explains why the IGNORE_DUP_KEY index option on a clustered index can result in much poorer performance than for a nonclustered unique index.
Itzik Ben-Gan reviews the logic and performance of some novel submissions to a reader challenge involving joining between two tables based on closest match.
Glenn Berry provides some details about how compatibility levels and the cardinality estimator have changed over the years, and why modern versions of SQL Server continue to be the better option.
