On Wed, 2003-04-30 at 11:42, dantrams wrote:
>
> Can you back this up with benchmarks (running IB/FB on Linux vs.
> Windows)?

Possibly when I have time, but even benchmarks can be disputed. There
will still be differences. If I was able to perform a bench mark it
would be

Java App using JayBird, on a Linux machine talking to a Linux FB server.

Which I then could do

Java App using JayBird on a Linux machine talking to a Windows FB
server.

I could also do the same with a C++ app and wrapper C++ library talking
to the Linux and Windows servers.

Now with that said. my Linux and windows FB server would have different
hardware that would further convolute the results. Not to mention my
Linux FB server running software RAID 1. My windows machine will not,
and I do not have the time to configure and build two identical machines
purely for benchmarking purposes.

Not unless someone wants to fund the effort. ;) Donations anyone. :)

When I have time I will slap together something. Keep in mind it will be
slapped together and there will be many differences to provoke further
dispute or skepticism.

However it should surprise people when they see a slower machine running
Linux stomp on a faster machine running Windows. Now with that said,
keep in mind my Linux machine is built, designed, and dedicated to
running as a Firebird DB server. It has no other duties and is dialed in
so Firebird can perform at it's best. No windows machine could come
close. The windows machine by default will be doing more than my Linux
machine. I will attempt to dial in and make the windows machine a good
FB sever. However I could never say a windows machine was a dedicated FB
server. Not with a GUI running, and other things as well.

> I see lots of claims like this but I've never seen an actual
> benchmark on
> similar hardware and similar settings (i.e Forced Writes). I'm really
> interested to see if Linux is indeed faster than Windows, everything
> else
> being the same.

Whether you believe the claims or not is up to you. All I can say, is
most of the are most likely true. For many obvious reasons.

There really can be no comparison of a Windows machine to Linux with

"everything else being the same"

Why?

Well Windows always runs a display.

No big deal just install and run X on Linux.

Well X does have more overhead and does not perform the same as a
display on Windows.

What windows has a faster display.

Yes, and no, but now were are way off topic.

So you can do your best to make both environments the same, but there
will always be differences. Each variable has an effect, and can be
disputed.

For me it was a no brainer. I hate MS. Even if Linux performed the same
or worse than Windows, I would possible still use it for many reasons.
Luckily Linux performs outstanding, and I have never even considered
using anything else.

My Firebird Dbs left the windows world long ago, never to return. Now
all Db development and deployment is done on Linux. I couldn't be
happier or more productive. Not to mention the savings on not having to
buy a Windows license. That equals another stick of ram or other piece
of hardware.

Although I do miss some of the windows tools, like Database Workbench
formerly Firebird/InterBase Workbench. Oh well luckily I have IBAccess.
It gets the job done.

--
Sincerely,
William L. Thomson Jr.
Support Group
Obsidian-Studios, Inc.
3548 Jamestown Ln.
Jacksonville, FL 32223
Phone/Fax 904.260.2445
http://www.obsidian-studios.com