I've got a Dell 2850 and a Sun v20z server with SLES8 x86/32-bit and
Oracle 9.2.0.7 installed.

The OS on each is the same SLES8 with automatic update run, both kernels
are 2.4.21-295-smp. The Dell hardware is 2x3Ghz/3Gig
RAM/PERC4DI/Mirrored 36Gig 15Krpm drives. The Sun hardware is
2x2.5Ghz/4gig/LSI-MPT/Mirrored 72gig 10Krpm drives. I've run unixbench
and the Sun scores significantly higher on the composite than the Dell.
I'm not a benchmark expert so I'm going to assume that the two machines
pretty much evenly matched in terms of overall performance.

Here's the problem: When I run a database create using Oracle dbca
generated scripts, the Dell creates the database in roughly 12 minutes,
the Sun box takes 48 minutes! The scripts and Oracle installs are
identical! The log output from both create scripts is identical except
for the timing. There are no errors in the create process.

So all things being equal or weighted toward the Sun box in terms of
performance, the Dell 2850 seems to be blowing away the AMD powered Sun
box in a simple Oracle database create test.

My only thought is that somehow when the Oracle binaries are compiled
that they generate optimizations for Intel chips and that those
optimizations don't work on AMD? Surely there is no way the 15K rpm
drives in the Dell are 4x faster than the 10K drives in the Sun. Simple
IO benchmarks with bonnie and dd show the sun IO is actually faster.

What the heck is going on? Anyone have any ideas?

Andy

--
To unsubscribe, email: suse-oracle-unsubscribe@(protected)
For additional commands, email: suse-oracle-help@(protected)
Please see http://www.suse.com/oracle/ before posting