In general, is there a performance difference between using table variables and using temporary tables in SQL Server 2000?
There can be a big performance difference between using table variables and temporary tables. In most cases, temporary tables are faster than table variables. I took the following tip from the private SQL Server MVP newsgroup and received permission from Microsoft to share it with you. One MVP noticed that although queries using table variables didn't generate parallel query plans on a large SMP box, similar queries using temporary tables (local or global) and running under the same circumstances did generate parallel plans.
A senior member of the SQL Server development team told me that table variables use internal metadata in a way that prevents the engine from using a table variable within a parallel query. He also said that SQL Server maintains statistics for queries that use temporary tables but not for queries that use table variables. Without statistics, SQL Server might choose a poor processing plan for a query that contains a table variable. The development team member added that you should limit your use of SQL Server 2000 table variables to reasonably small queries and data sets and use temporary tables for larger data sets.
This advice about table variables contradicts some past Microsoft information, which said that table variables were faster because they're created in memory. However, table variables can incur disk I/O in ways similar to temporary tables.
In deciding whether you should use table variables or temporary tables, benchmark both in your environment. I suspect that temporary tables will provide better performance on larger result sets in most cases. For more information about table variables, see the Microsoft article "INF: Frequently Asked Questions—SQL Server 2000—Table Variables" at http://support.microsoft.com/default.aspx?scid=kb;en-us;305977.
