Paul McBride gives another rant round on ISNULL() vs. COALESCE issue. He tested performance of nested ISNULL against COALESCE and it turned out that ISNULL is faster.
Almost year ago I did some performance tests on ISNULL vs. COALESCE issue in terms of investigating search with optional parameters in T-SQL. It is interesting that such a small problem has started a wave of discussions and tests on T-SQL related weblogs.
I think that ISNULL is faster than COALESCE just because COALESCE obviously uses ISNULL inside itself. But the performance should be rather similar due to optimizations and such. Just test it on the real environment.
Thursday, March 17, 2005 10:54 PM