[cvsnt] Re: rough benchmarks...

[David Somers]

John Cole wrote:

[snip]
> Our benchmarking was very simple, do a software build which checks out the
> code, builds, commits and tags.  We only counted the time for the
> checkout, commit and tag.
> 
> We expected linux (in this case Fedora Core 4) to win, but were surprised
> to
> have W2K3 server post the best numbers.  Here are our results:
[snip]
> Any suggestions to why our assumptions were wrong?  We used the latest RPM
> and MSI installs.
> 
> While not a valid benchmark it did have interesting results.

It would have been interesting to see how the reiserFS would have performed.

Also, it is not clear whether your benchmarks are for one client or many
clients running the tests simultaneously... the scalability as more clients
stress the server may be different depending on the platform (so Windows
may initially appear to be better, but Linux may scale better. YMMV).

As always, there are lies, damm lies, and benchmarks.

-- 
David Somers
PGP Key = 7E613D4E
Fingerprint = 53A0 D84B 7F90 F227 2EAB  4FD7 6278 E2A8 7E61 3D4E