Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Growth of Local Partitioned Bitmap Indexes

Re: Growth of Local Partitioned Bitmap Indexes

From: Billy Verreynne <vslabs_at_onwe.co.za>
Date: 7 Jan 2004 22:35:18 -0800
Message-ID: <1a75df45.0401072235.64ca1b02@posting.google.com>


"Jonathan Lewis" <jonathan_at_jlcomp.demon.co.uk> wrote

> Notes in-line.

<snipped>

Thanks Jonathan for the input. The storage mechanism used by bitmap indexes is now a tad clearer. Feels like parts of the Oracle docs are written in a strang English dialect that's at times hard for me to understand the actual details. :-)

Oh yeah - neglect to mention that this is a batch database and not really OLTP. These specific tables are hit by a single large transaction process - very little concurrent processing. Which is in part why I hoped to get away with it.

The index growth has become larger in recent months - recent production processing changes (and a few physical design changes, and an 8i to 9i upgrade) have been made.., including now using more status flags before the data is settled into the final 2 states at the end of the month.

Performance has however improved overall with a factor of 2 if not more (mostly due to physical re-design and partitioning). But seems like we now have reached a stage where the space required for the bitmap indexes during monthly processing are too costly versus the performance gains obtained it.

Besides, the reporting processes (which find the bitmap indexes most useful) usually run towards month-end anyway - thus that will likely be a much better time to slap these indexes on the table when the data has settled.

--
Billy
Received on Thu Jan 08 2004 - 00:35:18 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US