Re: Which Oracle book was this?

From: Michael Austin <maustin_at_firstdbasource.com>
Date: Fri, 24 Oct 2008 23:29:41 -0500
Message-ID: <8fxMk.4759$c45.1491@nlpi065.nbdc.sbc.com>


dba_222_at_yahoo.com wrote:
> Years ago, I skimmed an Oracle book at the bookstore.
> I'm sure it was written by Tom Kyte.
>
> In it, he critiqued a system that would store data in
> rows, instead of simple columns!
> One criticism being that it performed very badly.
>
>
> So, instead of a typical table structure:
>
> field1
> field2
> field3
> field4
>
>
> There was a lookup table with field codes, and what they meant.
>
> Fieldcode_id
> Fieldcode_name
>
>
> The "data" table would be structured:
>
> Fieldcode_id
> value
> etc.
>
> Doing simple queries was of course much more complex.
>
> I remember reading words to the effect:
> "How did it perform? Horribly, pathetically, awfully!"
>
>
> I'm unfortunately dealing with such a system now,
> and would like to show this critique to the brass.
>
>
> Which book can I find this in please?
>
> Thanks a lot!
>
>

I was aware of at least one system - for a name-brand pharmaceutical company that used this design. How did this one perform? It was running DEC now Oracle/Rdb (OpenVMS) and when tuned properly really screamed. Yes, the queries were very ugly, but when properly configured

  • did very well.

One benefit was it flexibility in self-describing the contents. When I first arrived at the site and saw this design, I was very skeptical as to how it would perform. Using MIXED Storage areas (loosely analogous to a IOT where index and the data for that index reside in the same data page (extent in Oracle terms)) performance was more than adequate. Received on Fri Oct 24 2008 - 23:29:41 CDT

Original text of this message