Re: Database Performance Problem between 3:00PM and 4:00PM

On Oct 23, 1:37 pm, zigzag..._at_yahoo.com wrote:
>
> Top wait events between 2:30PM-3:00PM
> Event Waits Timeouts Time (s)
> (ms) /txn
> ---------------------------- ------------ ---------- ---------- ------
> --------
> db file scattered read 18,476 0 66
> 4 342.1
> db file sequential read 3,708 0 5
> 1 68.7

> Top wait events between 3:00PM-3:30PM
>
> Event Waits Timeouts Time (s)
> (ms) /txn
> ---------------------------- ------------ ---------- ---------- ------
> --------
> db file scattered read 57,418 0 959
> 17 368.1
> db file sequential read 9,694 0 107
> 11 62.1

Yeah, like Robert said, from Oracle's viewpoint I/O is much slower during the bad time. There's 3X as many waits, taking 15-20X longer.

Sounds like they are rebuilding a mirror, or just pounding on something between your server and the spinning rust. Don't forget something silly like controller pathing on the SAN - someone might think they are not bothering you because they are going to a different set of disks, not realizing they are choking you out. Same with filling up some cache. Someone needs to go through your entire organization and find out all the things that happen at 3:00PM.

jg

--
@home.com is bogus.
http://arstechnica.com/news.ars/post/20071023-would-be-identity-thief-finds-himself-stumped-without-printer-drivers.html