The CIFSbench package
effugas at best.com
Tue Mar 9 21:43:51 GMT 1999
>Indeed. That's why there was a *MASSIVE* argument
>about CIFSBENCH at the last CIFS conference.
>I have *NO* confidence in a closed source code
>benchmark. And also *NO* confidence in a Win32
>based one also.
Closed source, you've got a point. I don't know about Win32.
Jeremy, feel free to correct me on this but the ideal benchmark isn't merely
CIFS but actual true Windows performance. I say this because "pie in the
sky" benchmarks(MIPS with NOPs) are meaningless and reduce overall
credibility. I offer the following as required for a testing suite:
1) Different benchmarks for 95/98 vs. NT. They read differently, they
score differently, they're different OS's. You can't just lump their
figures together. However, the specific OS's need to be tested because
they're the most likely to be used at a client install. It may be nice to
have a combined stat, though, but only if it fulfills the next
2) Multiple computers simultaneously accessing. Important? I think so;
you don't want the additional economies or diseconomies from having all
messages being sent to the same node.
3) Messages should be routed in the same way the user will end up using
them, so *most probably* the clients should mount the SMB server as a drive.
I don't know if this has a speed impact, though.
Number three has a powerful impact--this means that we can fairly use
something as simple as a disk benchmarker and use it to test Samba! If we
can argue that most IT shops *don't* use apps with custom Win32 SMB calls,
rather they merely have a bunch of machines mount a drive and run apps off
of it, then it's fair and valid to benchmark that exact behavior.
Gimme feedback on this. I think we can actually devise something here.
>(Yes I know I tout Samba NetBench numbers. But
>that's *marketing*. I have *NO* confidence in
>Netbench either - I just don't want to see it
>displaced with something just as bad or maybe
>worse :-) :-).
That harms our credibility, but then it's marketing and marketers *HAVE* no
credibility by definition.
If someone *can't* tell me their solution sucks, I can't totally believe
them if they say their solution doesn't.
More information about the samba-technical