Re: Index Question

From: Jens Mayer <jmayer_at_ratundtat.com>
Date: Tue, 11 May 1999 15:09:11 GMT
Message-ID: <37384612.2291755@news.space.net>

On Mon, 10 May 1999 17:58:27 -0400, "Van Messner" <vmessner_at_netaxis.com> wrote:

>Under version 7.3:
>I have a 1 Gig table with 9 million rows. One column is a number(12). This
>column is not unique and there are about 500 rows for each different number.
>The column has a non-unique index which takes about 300M. The table is
>updated frequently.
>
>Is this index giving enough benefit to justify the additional 300 Meg? It's
>reasonably selective but the column on which it's based is a short number.
>Based on the size of the index it seems like the index values might be
>longer than the column values. Is that a problem?
>
>Thanks
>
>Van
>
>

Hi Van,

As far as I know, your index is so huge because Oracle uses about 22 Bytes to store a number-value. the length of the value has no influence on the internal size of a number. Is it important for you to store the values as number ? Key values may be stored as char(12) as well and this will take only about 12 Bytes...

Jens Mayer
Rat & Tat Beratungsgesllschaft mbH
Muehlenkamp 6c
22303 Hamburg Received on Tue May 11 1999 - 10:09:11 CDT