Re: Index Calculation Question

From: Marc Marchioli <marcm_at_computek.net>
Date: 1995/09/14
Message-ID: <439qt2$1v9_at_news.computek.net>#1/1


In article <437d66$dpm_at_aplinfo.jhuapl.edu>, stan_novinsky_at_jhuapl.edu says...
>
>The Oracle7 Server Administration guide (pg 8-37 for OpenVMS)
>shows an example of calculating the avg. index size for an index of
>a table. The book states
>
> Entry Header 2 bytes
> ROWID 6 bytes
> F Total len bytes of all columns that store 128 bytes
 or
> less...
> V Total len bytes of all columns that store 128 or
 more ...
> D Combined data space of all index columns
>
>
>The book tends to lead one to believe that the size is calculated on
>all columns of the table. Is it actually only the columns that are
>used as indexes or is it actually all columns ? (I think it is only
>the columns usedas indexes). This would tend to make the tablespace
>size somewhat smaller for theindexes that the tablespace size for the
>actual tables that will be indexed...
>
>Also, the Entry Header in the book states 2 bytes, but the example
>below it uses 1 byte..
>Is it 1 byte or 2 bytes ????
>
>Can someone please confirm these observations.
>
>Thanks
>stan_novinsky_at_jhuapl.edu
>
>*********************************************************************
>* System Configuration: *
>* *
>* VAXstation 4000/90 * *
>* *
>* OpenVMS 6.1 *
>* Oracle 7.1, Forms 4.0 *
>* 3 RZ28's *
>* 128 Meg Mem *
>*********************************************************************

This algorithm is used in my utils. The index size is based on thhe actual columns used in the index. I used the stated size for the block header not what was used in the example. Ifind the calculation for the index size can be off by as much as 100% depending on the distribution of the data and the definition of the index. If you want to use the utils see 'Free ORACLE Reports and Utilities for UNIX'. These will save you lots of time aand are very usefull. Be sure to download all 8 parts and stick them back together! Received on Thu Sep 14 1995 - 00:00:00 CEST

Original text of this message