We are considering using the partition feature for the major objects in a
DSS database, primarily for maintenance reasons in order to purge and
archive data each month. Since the partition key will be month, and most of
the queries use data in the predicate, this should provide some performance
benefits too. I am concerned about performance for queries which do not use
the date, and will be forced to scan local indexes for each partition - this
is fortunately a low percentage of the query load, but I would hate to have
this queries run with extremely poor performance. Does anyone have any
benchmarked or observed performance for this type of situation?
-RB
Received on Thu Sep 23 1999 - 15:01:19 CDT
