Re: DBA - clustered tables - WHY ? - WHO USES THIS ?

From: Danny Roosens <roosens_at_ibm.net>
Date: 1996/08/02
Message-ID: <4trr79$2bt0_at_news-s01.ny.us.ibm.net>#1/1


Since no one seems to answer my question (sniff) I went on looking elsewhere an came across this note, which I'd like to share with the rest of you out there

Further remarks are still welcome, ......

PS

I'm the first I see, answering his own questions......

roosens wrote:
>In the seven year I work with ORACLE , and this in several major (Belgian) companies,
>I never came across anyone using CLUSTERED TABLES.
>
>Could any point out the advantage- when you should use or more important when NOT to
>use them.
>How can you estimate the profit ?
>How can you measure your profit after applying them ?
>
>Who will let us join his experience ??
>
>
>Thanks.
>
>ir Danny Roosens
>Genesys - Belgium
>

Cluster

A means of storing related rows from different tables in the same block.

Clustering data is a technique where the storage of records of related tables is intermixed, with
the dependent records of one table in the same block as those of the master record.

The logic behind this is that you probably most use the dependent records after you have accessed
the master table. Thats fine, but having a table in a cluster prevents you from dropping one and
not the other.

Also, the master and dependent records of other tables must fit into an Oracle block, typically 2k
(though adjustable at install time by a brave DBA). Now, if you are thinking of cluster
transactions with an account master record, how long before you have blown 2k ?

No Oracle DBA I ever knew, apart from Oracle staff, thought that using clusters was worth the
hassle. The only clusters in my databases are the internal ones. Received on Fri Aug 02 1996 - 00:00:00 CEST

Original text of this message