The use of a unique clustered index is an interesting solution, however, it requires 1.2 times the space currently consumed by the table and rewrites the table. IMHO, unless we are talking a major duplication i.e. 90% dupes, creating the clustered index will be slower.
Chuck
[EMAIL PROTECTED] wrote:
If you are using Sybase, you can delete duplicate rows from a table by creating a unique clustered index 'with ignore_dup_row' (and then dropping it again if you don't want to ignore later duplicate inserts). I haven't benchmarked but I think this will be somewhat faster than a stored procedure.
Anyway, this gets offtopic for the DBI list - if you want to write some SQL at the client for it, ask an SQL mailing list, and if you want to get the best performance on a particular RDBMS, ask a list for that system.
