On Monday 18 April 2005 12:28, Matthew Stringer wrote:
I have a client who wishes to benchmark 4 itentical Compaq DL380's (dual Xeon64's) which run high volume Apache/SQL services.
Currently 2 of them run Suse 9.1 with ReiserFS and RAID5.
I've been asked to build the other 2 with Red Hat Enterprise 4 (ext3), perform the comparison and then upgrade the Suse installs to Suse Enterprise 9 and repeat the test.
I'm assuming you can upgrade to Enterprise without having to rebuild the server, can you downgrade again or will that require a rebuild?
I believe that this excercise will be pointless as I'd be suprised is there would be much of a difference but the client is paying for this so I can't complain.
I've always thought that from a technology point of view there's little between normal and Enterprise versions of Linux.
Yes, but in this case I think you should look at a larger picture than strictly technology and strictly now. SUSE 9.1 will get out of maintenance in 1 year. Upgrading all the 30 servers won't be so pleasant. SLES will be maintained for much longer.
Other than differences between fily system types I can't see why there would be a measurable difference between the perfomance of RedHat and Suse as they're built on the same Kernel, this test has come about due to my preference of Linux distro. Which ever system performs the best its' OS will be used to build a whole new platform based on 30 machines.
The differences in performance shouldn't be significant at all. It's still Linux, after all, be it SUSE or Red Hat. But again, this is not about just performance. Having 30 servers, you'd definitely want a "mass administration" tool, to do stuff from a console on all the servers at once. That tool is ZENworks.
Perhaps I should sabotage the Red Hat builds to ensure the platform uses Suse!
Falling over to the dark side of the Force is easy :-) I'd say keep the play fair. Don't hit Red Hat, they're not the enemy.