Re: 1 Billion 11 Byte Words... Need to Check Uniqueness UsingOracle

Bryan W. Taylor doodled thusly:

>> I'm also not convinced that the Oracle sort algorithm is very efficient.
>
>What are you basing this on? Sorting is a core competency for a
>database application, so I'd be astounded if they don't do this pretty
>well. I'm assuming that the data in question here is randomly chosen

Well, you see: that was never a major concern with RDBMSs until sizes started getting very, very large.

So large that the uber argument of "hardware is cheap, just buy more" to solve performance design issues didn't apply any longer.

Many years ago in non RDBMS days, relative sort performance was a BIG issue. It's coming back again. About time.

I believe there is something in 9i to address that. As in: you can plug-in your own super-duper sort code if you want.

I also think that while the max speed for serial sort algorithms is well known, for parallel sorts things are a little different. Maybe that's what Keith was referring to with the 10X factor?

Cheers
Nuno Souto
nsouto_at_optushome.com.au.nospam Received on Mon Feb 11 2002 - 04:20:04 CST