Re: HP or Sun for Oracle

From: Paul Beardsell <psb_at_sambusys.demon.co.uk>
Date: Tue, 18 Oct 1994 14:23:16 +0000
Message-ID: <781852347snz_at_sambusys.demon.co.uk>


In article <37b0dg$rlf_at_uk-usenet.uk.sun.com>

           timr_at_cerrera.uk.sun.com
           "Tim Read - Sun Linlithgow - Principal SE and DB" writes:

> ----> Paul Beardsell
> And I'm interested: what legitimate reason could Oracle have to show
> Sun its source code? And vice versa. Do they do out of a spirit of
> openness?
>
> <----
>
> So that both parties can tune their products w.r.t the other and thereby
> optimise the overall performance.
>
> Tim
> ---

I'm intrigued: Is this simply the first reason that pops into your head or, as a Sun employee, do you yourself have reason to examine Oracle source.

The reasoning behind my scepticism is as follows: Oracle, to be portable, will be using POSIX calls (almost?) exclusively. Sun will provide all of the POSIX interface and should _already_ be concentrating on the performance of each of them. There is little that Sun can do to improve the performance of one POSIX call at the _expense_ of the performance of another. So they should just get ahead and do the tuning enhancements without reference Oracle.

Consider also the characteristics of the Oracle application from the point of view of the Operating System. I'm not suggesting that Oracle is not a sophisticated, complicated product but from the point of view of the OS it is simply reading and writing bytes to and from files and to and from the network and to and from shared memory. There is _nothing_ unusual about Oracle in comparison to other applications when looked at at this low level.

For instance, there is nothing that Sun can do to its OS improve the performance of an Oracle sort that would require access to Oracle source. Any improvement to the execution of the compare or move opcodes would favour Sybase just as much as Oracle. Similarly with any change to the kernel's lseek() routine.

Consider also: What can Sun tell Oracle about its product that will not also make Oracle run quicker on another Unix?

If Oracle and Sun do have access to each other's source then I guess that it of little use to either party. More than likely, IN(H?)O, that this mutual access to source is more use as a marketing tool than anywhere else. My guess is that at a meeting between Oracle and Sun senior managers they agreed that it would be a "good idea" to share source code.

-- 
Paul Beardsell                          SAM Business Systems Ltd
~~~~~~~~~~~~~~                          21 Finn House, Bevenden St,
pbeardsell_at_cix.compulink.co.uk          London, N1-6BN, UK.
psb_at_sambusys.demon.co.uk                (+44 or 0)71 608-2391
Received on Tue Oct 18 1994 - 15:23:16 CET

Original text of this message