Hi,

I have a couple of tables with at least 5 million rows each from which ad-hoc reports generated. The reports can be sliced-and-diced using a combination of any of the 15+ columns on these tables. My goal is to improve the poor performance of these reports. Questions:

1- Do you advice that I create an index on each of the columns used in the where clause (taking into account all the various scenarios that the report can be sliced-and-diced)
2. What kind of index - bitmap or b-tree. The environment is a dss (data warehouse) and the average # of distinct rows is 3800 out of 5 million rows.
3. If the tables are partitioned (by range), is it recommended to also partition the indexes (using the same range as the table). Comments are also welcome.


Thanks,
L