Re: 9i multi cache buffer

"Jonathan Lewis" <jonathan_at_jlcomp.demon.co.uk> wrote in message news:ar2gi0$n7d$1$8300dec7_at_news.demon.co.uk...
>
> It was more of a hypothetical question than an
> absolute need for information. And Nuno's comment
> on this issue was also relevant - different machines
> have difference memory page sizes, and can optimise
> some actions based on memory paging.
>
> Many years ago, an HP specialist pointed out to me
> that the HP had a default memory page size of 4K,
> and suggested that it's filesystem handling could therefore
> be affected by choosing a 4K block filesystem rather than
> an 8K filesystem. It isn't the sort of thing you can test
> realistically, unfortunately - in principle it sounds correct,
> and in principle it sounds as if you just set up lots of
> different file systems for lots of different Oracle block sizes.
>
>
> This means, of course, that I have to disagree with (at
> least) some of your response to Geoff Ingram - there are
> arguments for choosing to maintain different block sizes
> in the databases that do not relate to the speed of the
> file system, and even if the file-system interferes in some
> way, there are still cases where other considerations can
> be more important. However, I suspect that many of those
> arguments will never be subject to a strict test.
>

I don't disagree. I guess I start out from a different angle... namely, that odd blocksizes are in *addition* to the standard one. That it's therefore important (still) to get the right standard blocksize. So I bang on about the initial decision so much because I'd rather people didn't get it wrong. I also suspect that many (maybe most) decisions to add in extra blocksizes are made for the wrong reasons: it's now the "easy" thing to do, instead of mucking around with other things such as PCTFREE or MAXTRANS first.

It's very difficult to teach greys on beginner courses. That's why I stick to black and white as far as possible, and leave the greys to the advanced ones. You can always add 'subtle' on top of 'sound principles' later. You start off with dodgy principles, and there's not much anyone can do about it afterwards!

Regards
HJR
>
> --
> Regards
>
> Jonathan Lewis
> http://www.jlcomp.demon.co.uk
>
> Next Seminar dates:
> (see http://www.jlcomp.demon.co.uk/seminar.html )
>
> ____USA__________November 7/9 (Detroit)
> ____USA__________November 19/21 (Dallas)
> ____England______November 12/14
>
> The Co-operative Oracle Users' FAQ
> http://www.jlcomp.demon.co.uk/faq/ind_faq.html
>
>
>
>
>
> Howard J. Rogers wrote in message
> <7JhA9.75556$g9.212743_at_newsfeeds.bigpond.com>...
> >You're asking me????
> >
> >My advice on such matters is (1) consult your vendor and (2) consult
> Steve
> >Adams, who says that HP-UX has an 8K block size. He goes on to say
> the
> >following (which makes about as much sense to me as the output from
> top):
> >
> >If your operating system is not listed in the table above, you should
> be
> >able to work out the file system buffer size by studying the ratio of
> blks/s
> >to r+w/s in the sar -d output for various buffered filesystems.
> >
> >Let us know what the outcome is, so I can shut up about a 'right'
> block
> >size.
> >
> >Regards
> >HJR
> >
>
>
>
>
>
Received on Fri Nov 15 2002 - 13:06:48 CST