Re: data dictionary with a million tables

Luis,
  as well as look into the technical side you or the developers   should look at the logical decelopment side.

  1 Millions tables would suggest that you have 1 Millions of    different entities in your business environment (ok ... I know    that is how it should be in an ideal 3rd denormalised world),    it just doesn't look realisitic to me.

  From the maintances point of few ... 1 Million tables will have   at least one primary key each and about 50% will be connected   via FK, some of them will be using check constraints and maybe   some other indexes (every index represents an entry point from   the application) ... you will see ... you will be reaching   easily the 5 Million limit.

  Who the hell will ... and how .. maintain this enviroment,     and more who ... will pay for it ("who will pay for it" is     normally a question which wakes people up, as nobody     wants to pay).

  Important would be to emphasize that the cost of a software    is not restricted to the development of software, an important    factor is as well the cost for maintainance.

  I would recomment to take a break from create the tables, sit   back and go back to the logical DB design. You might consider   to user Oracle Designer/2000 (an investment of a few bucks) in   order to support your development with their ideas.

  Also ... I realise that 3rd de-normalisation is a goal to     achieve, but sometimes it makes more sense (eg. performance)     no to do so (ok ... with 1000 rows each it wouldn't matter     much)

  Something different: highly denormalised environments usally     benefits from use of clusters.
    Also ... lot of these tables will have just a couple of

     columns, specially the intersection tables (resultion of
     N:M relationsships) which just contain FKs are primary
     candiadates for index organised tables.

Let us know if you need further help
Regards
Dante
In article <3730EA76.15F080_at_bigplanet.net>,   Luis Londono <luisl_at_bigplanet.net> wrote:
> I need help with a littlle problem that I have. I have a bunch of
> developers that want to create an application using Oracle that would
> end up having probably a million very small tables (100 - 1000 records
> each). It is easier for them to develop it this way, but I am trying to
> convince them that it is not a good idea. It seems to me that the
> performance of the entire database would suffer a lot since the data
> dictionary would be over worked. Unfortunately I do not have any solid
> evidence or an argument more convincing than a gut feeling. Does anyone
> have some information that would help me? Or if I am wrong, I would
> sure like to know. Thanks.
>
> -Luis
>
> --
> Luis Londono
> Senior Systems Engineer 370 E 1130 S
> Big Planet Inc. Provo, UT 84606
> mailto: luisl_at_bigplanet.net (801) 345-7222
>
>

-----------== Posted via Deja News, The Discussion Network ==---------- http://www.dejanews.com/ Search, Read, Discuss, or Start Your Own Received on Sat May 08 1999 - 04:22:57 CDT