Re: parallel Query

And ..

What's the version of Oracle ?
What are the v$pq_tqstats on completion ? Which process's trace files are you running tkprof against ? Is the whole fact required by the join, or can t.t_id be null ? Quantify 'serious' degradation

--

Jonathan Lewis
Yet another Oracle-related web site:  http://www.jlcomp.demon.co.uk

sergey_s_at_my-deja.com wrote in message <8nknll$b66$1_at_nnrp1.deja.com>...


>Hard to tell by looking at this. Post the execution plans. How big are

>the two tables? How much memory do you have? What's the db block size?

>What's the OS?

>

>

>Sergey

>

>In article <8nkb0b$e19$2_at_neptunium.btinternet.com>,

>  "Dean" <d.j.l_at_btinternet.com> wrote:

>> Hi,

>> can any shed some light on the following.

>>

>> PQ seems to work fine on fact table. ie

>> select sum(sales), t_id from fact group by t_id

>> PQ gives me 4 fold increase in performance.

>> However..

>> select sum(f.sales), d.t_id from fact f, dimension d

>> where f.t_id=d.t_id

>>  group by d.t_id

>>

>> this leads to serious degradition with shed-loads of io (which is'nt

>> reported in tkprof stats)

>> Oracle support have had this tar going for 2 months with no joy

 (quelle


>> surprise)

>>

>> any help is gratefully appreciated

>>

>> Cheers

>>

>> Dean

>>

>>

>

>

>Sent via Deja.com http://www.deja.com/

>Before you buy.