Re: Database Hit Ratios

In article <miNW8.31692$Hj3.95941_at_newsfeeds.bigpond.com>, you said (and I quote):
>
> For many, it's all obvious but for some it's a not so clear.

Yup. The problem with all these ratios is not their calculation or use by themselves. It's the interpretations people give to certain values and the myths created around some "magic recipes".

You know how it works:

Someone writes in a book or paper that "I got BHR from 98% to 98.1% and performance picked up for my application".

Two weeks later it becomes urban legend that if you "don't keep BHR at 98.1% your application performance will never pick up".

And so many others. The fragmentation one is a classic that still lingers today.

Probably very well intentioned when they first came out, but progressively distorted and modified by constant use out of context. To the point where they are much more damaging than helpful.

The cure? I dunno. Not certified to talk about it... ;-) However, I'd rather understand Jonathan's and Steve's stuff about performance than apply ANY ratios. A personal choice maybe, but one I'll stick with for a while.

I know it will eventually land me into hot water tough: first time I come up against Oracle's consultancy services and they complain to a client about some obscure out of context ratio I couldn't be bothered even to look at, who do you reckon is gonna win the argument? Not me for sure. It's happened in the past and it will happen again. C'est la vie.

Moral of the story? Follow what you think is right but make sure your derriere is covered.

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