| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Usenet -> c.d.o.server -> Re: tkprof question
Richard
Like yourself I've found it hard to measure much if any impact of using timed stats. I've always quoted 2-3% from early days of Oracle 7 when people used to quote 5% at me and I always found it was less. I think in most cases I found it less than 1% but hedging my bets always seemed to give dba's a warm glow of contentment when things were better than predicted :-) whilst still retaining some self respect by quoting 2-3% **tops** :-)). I haven't measured it recently - it's not something I've had to justify turning on - I've bigger battles to fight :-(.
I always use timed stats as I go on the assumption that the benefits (in better tuning) outweight the, if measurable, performance loss from using it. It sounds like I'm "lucky" that I'm still trying to convince our Tech director that shared sql is beneficial - we don't use it yet, such that I've managed to avoid the nastiness described by Norman (we're still on 8.1.7.0 !!! just please don't ask why !)
Andy
"Richard Foote" <richard.foote_at_bigpond.com> wrote in message
news:xo1F9.84819$g9.239538_at_newsfeeds.bigpond.com...
> "Andy" <andy.spaven_at_eps-hq.co.uk> wrote in message
> news:qW%E9.2641$9R.10226819_at_newsr2.u-net.net...
> > To get CPU timings you need the initialization parameter
timed_statistics
> > set to TRUE for either the whole db or just your session. This can be
> done
> > by
> >
> > 1) permanently in the init.ora file (see your dba) or
> > 2) by issuing alter system set timed_statistics=true in a sql session
(see
> > your dba or risk his wrath altering system parameters).
> > 3) by issuing alter session set timed_statistics=true in a sql session
> (not
> > that you're likely to be able to do this for a third party product
unless
> > you have the source and can rebuild it - yuk).
> >
> > 1 & 2 will have an impact on performance as Oracle gathers statistics
for
> > all sessions but only a small one (2-3% tops).
> >
>
>
>
 |
 |