Re: What is Parallel DML?

"Billy Verreynne" <vslabs_at_onwe.co.za> wrote in message news:9jlp21$990$1_at_ctb-nnrp1.saix.net...
> "Howard J. Rogers" <howardjr_at_www.com> wrote
>
> > Read Jonathan Lewis' post on the subject, and you will see that
> > my argument is not nonsensical, nor wrong-headed, nor circular.
>
> Jon simply described the pros and cons of PQ. As I recall, you
 specifically said
> that PQ is pointless on a single CPU platform.

You'll understand, I'm sure, if I take the rest of your post as fundamentally agreeing with what I *actually* said, which was that it was pointless on a single CPU box, really, but worked nicely with multiple CPUs.

Most people are not going to be working in a Parallel Server environment, using MIPS processors. Rather more people *are* going to be running in dedicated server mode on Intel chips.

Apparently there you found a degree of parallelism of 2 able to cripple your machine.

I rest my case.

Jonathan's post was not just a description of the pros and cons of PQ: it explained very nicely that whilst you *might* get PQ producing benefits on a single CPU box, *it would all depend*. It requires thought, and testing, and measurement as to whether to enable it at all. Rather less thought about that particular decision would be needed on a multi-CPU box (though the degree of paralellism to use would still want watching).

HJR
>And that is what I've taken
> exception too as that concept _is_ an old wife's tale that rears its head
 in
> this newsgroup every so often.
>
> > Arrogant shits who refuse to look facts in the face annoy me.
>
> I may come across arrogant Howard, but PQ is one thing about Oracle that I
 do
> know something about. And unlike you, my experience is not from a couple
 of
> simplistic experiments, but from a couple of years looking after an OPS
> warehouse on a MPP cluster where each cluster node was a single CPU box..
 and we
> were running PQ on each node. To get technical - back then, each node was
 a
> single MIPS R4000 200Mhz CPU running Reliant Unix and Oracle 7.3 (and then
 7.4).
> I found that 8 to 10 PQ's per node gave me the best CPU utilisation.
 Unlike on a
> Intel MMX 200Mhz NT platform running Oracle 7.3 (or was it 7.1) and NT
 3.5 - my
> state-of-the-art little R&D box at time, where anything more than 2 to 3
 PQ's
> started to throttle the CPU. But even there, using 2 PQ's doing a FTS
 resulted
> in significantly faster performance, than using a single process instead.
>
> PQ pointless on a single CPU platform? Think again.
>
>
> --
> Billy
>
>
>
>
Received on Wed Jul 25 2001 - 15:49:07 CDT