Data Skew

I have the following situation -

Table A has 10 rows and the primary key is an sequence generated id. The only other column is a 'description' field.

Table B has 10 million records and has an indexed foreign key pointing to the table A pk.

The foreign key values in the table B are massively skewed. 95% of the records have a single value. And any queries against these tables can only use the 'description' column in Table A to limit the records produced in the join to table B.

So histograms on the fk are not effective since the limit is on the description column. How can I get the optimizer to know the optimal access strategy for table B (based on the skew) if the queries can only be limited on table A.description?

Thanks. Received on Fri Apr 15 2005 - 15:28:52 CDT