Re: bitmap index and distinct values

From: Jonathan Lewis <jonathan_at_jlcomp.demon.co.uk>
Date: Sun, 26 Apr 2015 11:11:02 +0100
Message-ID: <7AC06B14A0294F3181ECC9A14E2763EC_at_Primary>


CP,

Your array is presumably a two-dimensional array where each row represents a possible value, and each column represents a row in the table.

If so, then imagine a B-tree which is built from just the first row where each distinct value appears, then attach to each entry of that B-tree the row from your array corresponding to that value.

We then have to add a little complication because (a) we eliminate lots of consecutive zeros from the row we store, and (b) the row could (still) be much too large for a single B-tree entry so we cut the row into chunks and store each chunk as a separate b-tree entry, so each entry actually looks like: (value, start_rowid_of_chunk, end_rowid_of_chunk, chunk).

If you've read Richard Foote's material you've probably had enough of textual descriptions, but the first of the three items on this page may give you just a little nudge to seeing how it works: http://jonathanlewis.wordpress.com/2015/04/23/golden-oldies

Regards

Jonathan Lewis
http://jonathanlewis.wordpress.com/all-postings

Author: Oracle Core (Apress 2011)
http://www.apress.com/9781430239543

  • Original Message ----- | | Am 25.04.2015 um 20:05 schrieb Cee Pee:
    | > Thank you Martin and Ram. I read the link, it is bit too technical for
    | > me to follow. A picture would help a lot. I have seen pictures related
    | > to btree index structure,but none on bitmap index structure. Do you
    | > happen to have a diagram related to bitmap index. I can compare the
    | > picture and the article. The picture of bitmap indexes in my mind are
    | > more of an array shaped structures, not a tree shaped as the article
    | > seems to suggest. thanks.
    | >


No virus found in this message.
Checked by AVG - www.avg.com
Version: 2015.0.5863 / Virus Database: 4334/9622 - Release Date: 04/25/15
--
http://www.freelists.org/webpage/oracle-l
Received on Sun Apr 26 2015 - 12:11:02 CEST

Original text of this message