Re: Cache Hit Ratio from system views

On Sep 5, 9:33 pm, Noons <wizofo..._at_yahoo.com.au> wrote:
> On Sep 3, 7:19 pm, Niall Litchfield <niall.litchfi..._at_gmail.com>
> wrote:
>
>
>
> > > BHCR shows disk i/o percentage. An important indicator for buffer cache
> > > tuning.
>
> > So how exactly would you use it? If BCHR falls below a certain value
> > increase one or more of the pool sizes? If it rises above a certain
> > value reduce them?
>
> I think everyone knows where I stand on this subject.
> However, I'd just like to point out that these words:
> "The buffer cache hit ratio can be used to verify the physical I/O as
> predicted by V$DB_CACHE_ADVICE."
> are even as we write in the relevant section of the performance tuning
> guide of oracle 10g, specifically in here:http://download.oracle.com/docs/cd/B19306_01/server.102/b14211/memory...
>
> Pity this thing is to this day still promoted as a relevant
> performance
> measurement tool. Now: if this was v6 and there was nothing else
> available, I can understand why we'd still talk about it.
> Nowadays? Let's move on, much better stuff to look at.
>

I think at one point I noted this, as well as some v$bh formulae in the perf manual, and not long after Jonathan wrote something about advice on usenet :-)

Personally, I've found the cache advice useful as a gross ballpark measure, as long as you don't consider it godlike advice. It you did iteratively follow it, you'd keep on increasing the cache forever.

The performance manual has gotten much better over the years, but still doesn't get anywhere near the top books and articles, like Jonathan's post about this thread.

> > shouldn't be news given that it was first described and demonstrated
> > by Knuth when I was 4 (and I'm ten times that age this year). The more
>
> hippo birdie in advance! or late? Anyways: big 4-0, eh? Congrats!

Scare off the two ewes, noons? :-)

jg

--
@home.com is bogus.
Happy Birthday to Roger Waters!
(Special "What Was I Thinking?" award for the leaving Pink Floyd
thing).