Re: Got the darn buffer busy waits under control, at last!

In article <3D0F3C42.6D1B3F46_at_more.net>, you said (and I quote):
>
> Sounds like a contradiction to me. Do they perceive improvement because
> of the spin count tuning or not?

Ricky, the users wouldn't have a clue if spincount was changed or even when. What I do nowadays is tell them that nothing when there is a change. Then a few days (before or after, depending on my mood), I tell them there was a major change made. Immediately all sorts of "problems" crop up from the usual suspects. They are invariably the ones that have been trying to push their own cart when it comes to Oracle administration and/or tuning.

It may sound cynical, but I'm getting too old for this game... ;-)

>
> Ratio tuning like this is nonsense, of course. I think you already know
> this.

Yes. But try to explain that to folks sold on Toad being the be-all-and- end-all of performance data collection and tuning information...

> The process run queue, averaged per cpu, will tell you right away if you
> are cpu-bound. Similar information is offered for IO.

Yes, I've found the I/O queue lengths are actually much more relevant than the individual disk access times calculated by sar.

> The ratios don't really illustrate this. And, don't forget to evaluate
> variance - don't just peek at a single snap. Plus, the report must be
> from the same busy period when the "problem" occurs. Lots of people fail
> to follow these rules and end up with garbage.

Oh yes, I'm a very big fan of sifting through /var/adm/sa/sa*! ;-) One of those things that gets me is all those guys with sar or glance running interactive with the default loop time of 1 or 5 seconds. Absolutely USELESS for any serious tuning purposes! Yet, I see them everyday...

>
> There is a new white paper on Metalink that describes this method and
> illustrates operating system performance analysis. Look under "top tech
> docs", "performance", "database", "tuning methodology..." and it should
> be about the first one listed. COE Performance Method.
>

Thanks muchly. Will have a look at it.

-- 
Cheers
Nuno Souto
nsouto_at_optushome.com.au.nospam