Re: Index cluster with one table

From: Jonathan Lewis <jonathan_at_jlcomp.demon.co.uk>
Date: Fri, 25 May 2012 13:44:24 +0100
Message-ID: <t8udnSzS79UA4yLSnZ2dnUVZ7sudnZ2d_at_bt.com>


"vsevolod afanassiev" <vsevolod.afanassiev_at_gmail.com> wrote in message news:75f08b8f-7a5d-419e-b0db-440830e46ed8_at_t2g2000pbg.googlegroups.com...
|A vendor provided scripts for table creation in a new database.
| Some tables get created in index clusters, however each cluster
| contains only one table.
| Does it make sense? I thought that idea behind clusters was to combine
| two or more tables to make joins faster.
|

Yes, it can make sense.

You might want to cluster together rows from the same table even though they appear at very different times. Clustering for joins was just one reason for clustering. Note that Oracle uses many single table (hash) clusters in its TPC-C records.

The same time of thinking applies to Index Organized Tables - data arrives in one pattern, but is stored in another pattern for faster (group) retrieval.

-- 
Regards

Jonathan Lewis
http://jonathanlewis.wordpress.com
Oracle Core (Apress 2011)
http://www.apress.com/9781430239543
Received on Fri May 25 2012 - 07:44:24 CDT

Original text of this message