Performance of Oracle on read-intensive operations

From: Vsevolod Afanassiev <vafanassiev_at_westpac.com.au>
Date: Mon, 20 Sep 1999 16:20:39 +1000
Message-ID: <7s4joa$g8m3@xlprod01.westpac.com.au>

What should be performance of Oracle on read-intensive operations vs the performance of copy (UNX cp) command? For example, I have a table with 5 mln. records, each record is approx 100 bytes.
I do SELECT without any WHERE, ORDER BY, GROUP BY, or anything like that. I run this SELECT in SQL*Plus with all options (like LINESIZE, PAGESIZE, etc)
set appropriately. SQL*Plus creates spool size 0.5 GB This operation takes 20 min
on the box with very light load (all CPUs are 70% idle).

At the same time, to copy a 0.5GB file between filesystems only takes 2 min. Is it normal? Is it because Oracle needs to manipulate blocks in the buffer cache
(write dirty blocks / free space for news blocks)? Or these is some other reason?

Thanks,
Sev Received on Mon Sep 20 1999 - 01:20:39 CDT