Re: Oracle benchmark

<sybrandb_at_yahoo.com> ha scritto nel messaggio news:1141379982.512484.262190_at_z34g2000cwc.googlegroups.com...
> The best hint I can give you is to read the Oracle Concepts manual, and
> the Performance Tuning manual.

ok.

> You are wasting your time, as you don't seem to understand anything
> about how the buffer cache mechanism works. You are up to an exercise
> in futility, because Oracle never has the complete database in SGA.

I think the same thing but it's a thesis for my degree, my professor ask me to put all database in SGA, so if it's possible (but seems not) I have to do it...

> Even if it would have been possible, the figures you call 'benchmarks'
> would be completely irrelevant once the database starts growing.

The db is generate by 'tpc-h dbgen' and it don't grows. I have several query to execute, like this, in wich I have to eliminate I/O access to do an speed-up test. Test is about growing in speed when I increase number of fisical processors (Solaris and Sparc environment).

select

        l_returnflag,
        l_linestatus,
        sum(l_quantity) as sum_qty,
        sum(l_extendedprice) as sum_base_price,
        sum(l_extendedprice * (1 - l_discount)) as sum_disc_price,
        sum(l_extendedprice * (1 - l_discount) * (1 + l_tax)) as sum_charge,
        avg(l_quantity) as avg_qty,
        avg(l_extendedprice) as avg_price,
        avg(l_discount) as avg_disc,
        count(*) as count_order
from
        lineitem
where
        l_shipdate <= date '1998-12-01' - interval '62' day (3)
group by
        l_returnflag,
        l_linestatus
order by
        l_returnflag,
        l_linestatus;

I did many tests but never be able to stop I/O access. But if you say SGA never contains entire DB I belive it.
Tanks for the answer and forgive me for my bad english. Received on Fri Mar 03 2006 - 04:38:02 CST