|
Posted by Gert-Jan Strik on 11/26/06 20:35
ramraj@gmail.com wrote:
>
> Hello people,
> I might sound a little bit crazy, but is there any possibility that you
> can incorporate 4^15 (1,073,741,824) tables into a SQL Database?
>
> I mean, is it possible at all? There might be a question of where
> anyone would want so many tables, but i'm a bioinformatics guy and I'm
> trying to deal with genomic sequences and was coming up with a new
> algorithm, where the only limit is the number of tables I can put into
> a Database.
>
> So, can you please advise if its possible to put in so many tables into
> a SQL database? Or is the Bekerley DB better?
In addition to Erland's answer: there might be some practical
limitations.
The smallest possible table, without indexes and only one row will
reserve 16 kilobytes. If you were to create 4^15 tables (without
indexes, primary key and unique constraint), and populate each table
with only 1 row, then you would need 16 TB. If all these tables have a
primary key constraint, that would be 24 TB. Also, if only a few bytes
per 8KB-page are used, then the Buffer Pool cannot be managed
efficiently, with the risk of poor performance.
Gert-Jan
P.S. Yes, it does sound crazy. If at all possible, then don't do it...
[Back to original message]
|