Re: Statspack ratios help

well in theory yes but reality it can mean something entirely different. A
100% hit can mean that you are writing useless sql which floods the buffer
cache with loads of blocks and then you are naturally going to hit in in the
cahce next time round with another query

now to some people that would be nirvana, but the fact that you are doing
thousands and thousands of LIO's per query passes them, you still have poor
performacne even though the cache ratio is 100% (reason why you shouldnt
rely on it being a good indicator).

Of course it could be the opposite and you have the best application in the
world

Now in this case he says he is using hibernate to generate the sql, knowing
a bit about hibernate and the number of io's his statspack report, i'm
guessing it is a bit of a shocker

asktom explains this better than me obviously


On 06/06/06, Bryan Wells <bunjibry@xxxxxxxxx> wrote:

As a Jr. wanting to be Sr. (someday) I have to ask the stupid questions:

why is Buffer % bad?  Dont you want 100% buffer cache hit vs. 100% disk
i/o?

Other related posts: