Wierd benchmark results - please advise.
Date: Thu, 15 Nov 2001 14:55:42 +0000 (UTC)
Message-ID: <9t0l1e$r8t$1_at_news.huji.ac.il>
As part of my job I wrote a benchmark program to compare the performance
of DB2, Oracle, and Informix.
I wrote the program in Visual C, on MS-Windows 2000, using embedded SQL.
The heaviest part of the trade study, as it turns out, is filling a table
which has three blobs (20KB, 60KB, and 520KB) with 1,000 records.
Both Informix and Oracle took about 65 seconds to perform 1,000 inserts
into the table, with the BLOB data coming from RAM - three arrays of
unsigned chars filled with random data before each insert.
To my surprise DB2 did those inserts in about 6 seconds - an order of
magnitude faster than Oracle and Informix.
All the databases ran on the same machine (1GHz Pentium 3, 512 MB of RAM,
SCSI disk for the data seperate from the system disk), and I checked that
DB2 has indeed inserted the data into the tables. Representatives of all
databases vendors had the chance to optimize the database software, so it's
not that my lack of knowledge in database optimization gave DB2 some
advantage in a freak situation.
This result looks fishy to me.
Can anyone suggest an explanation to this result ?
TIA,
Uri Raz.
--
+----------+----------------------------+------------------------------+
| Uri Raz | s2845543_at_t2.technion.ac.il | kaum vit litz, tor lod vike. |
| TCP/IP Resources List http://www.private.org.il/tcpip_rl.html |
Received on Thu Nov 15 2001 - 15:55:42 CET