RE: Insert Rate/Second

From: mkb <mkb125_at_yahoo.com>
Date: Wed, 24 Aug 2005 11:59:55 -0700 (PDT)
Message-ID: <20050824185955.75199.qmail@web32801.mail.mud.yahoo.com>

• mkb <mkb125_at_yahoo.com> wrote:

> --- Bruce McCartney
> <bruce.mccartney_at_dbinfosystems.com> wrote:
>
> > Hi Mohammed,
> >
> > If you want an accurate reading by table you can
> use
> > dba_tab_modifications;
> > something like this, I used SYSTEM so had to grant
> > access:
> >
> > grant all on SYS.DBA_TAB_MODIFICATIONS to system
> > with grant option;
> >
> >-----------8<-------------
>
> Hi Bruce,

more snippage.....

--------------8<--------------------

> I tried it on a small test table by inserting about
> 7k
> records. Running the script, gives a result of
> about
> 230 inserts a second which doesn't seem right. It
> should about 7k rows/second as shown below:
>
> 18:15:23 SQL> /
>
> 7874 rows created.
>
> 18:15:24 SQL> commit;
>
> Maybe I should remove the conversion-to-seconds
> factor
> in the above SQL? But if I do, I would get insert
> rates in the millions/second. Now that don't seem
> right either.
>

No...wait, Bruce's script and idea is accurate if you also include the conversion-to-seconds factor, it was me who screwed up. When the script runs, it issues a dbms_lock.sleep(30). If you multiply the INS_RATE by 30, you'll get a figure close to 7k rows/sec.

In fact, changing sleep(30) to sleep(1) also gives a value close 7k rows/sec. Getting rid of it altogether, also works.

So what was the idea behind the sleep?

Thanks again.

--
mohammed


		
____________________________________________________
Start your day with Yahoo! - make it your home page 
http://www.yahoo.com/r/hs 
 
--
http://www.freelists.org/webpage/oracle-l

Received on Wed Aug 24 2005 - 14:02:00 CDT