Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Oracle 7.3 Very Large Tables

Re: Oracle 7.3 Very Large Tables

From: Vincent Birlouez <birlouezv_at_logica.com>
Date: Tue, 05 May 1998 11:10:21 +0100
Message-ID: <354EE58D.45E26F0D@logica.com>


I had a table with more than 50 millions rows on Oracle 7.3.3 and I did not experienced any problems with that.

The Server Reference Guide (5-2) says there is no limits in term of rows with Oracle 7.3 but there is a limit in term of blocks : not more than 2e32-1 blocks...

Keith Fulton wrote:

> I am currently working on an Oracle 7.3.4 application on a Sequent NUMA-Q with
> heavy batch TP requirements at night, and heavy reporting and inquiry during
> the day. Our most detailed tables are projected to grow approximately 9
> million rows / month, and we need to keep 12-18 months of history online for
> inquiry purposes.
>
> We have several DBAs who are saying that Oracle cannot support this
> requirement, and that Oracle really can't handle more than about 25 million
> rows in a table, but this seems much smaller than a lot of data warehouses and
> other big installations. So first I would like to know if anyone out there
> has guidelines on just how big tables can get before Oracle will "break" or
> non-linearly degrade performance.
>
> Those same DBAs recommended that we use O7.3 table partitioning with views to
> accomplish the scalability we need, and (not to my surprise) that was a factor
> of 10 slower than just the big table. Now they are saying the only solution
> is to go to Oracle8, which handles partitioning much better (but which I
> think is still not mature enough).
>
> Another idea we had to accomplish our scability and history needs are to split
> up these tables into 2 parts--a) the 'current' data for say the last 60 days,
> and b) historical data only there for inquiry purposes. This solution, in
> effect, splits the table into the TP and data warehouse pieces, and seems
> better, but leaves the data warehouse piece at 100+ million rows.
>
> Any feedback on this sort of problem would be greatly appreciated.
>
> KWF
Received on Tue May 05 1998 - 05:10:21 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US