Re: FTP benchmark proposal

Larry McVoy (lm@bitmover.com)
Mon, 28 Jun 1999 11:13:40 -0600

: > We are talking about an ftp benchmark here to see if we can actually
: > keep up with the FreeBSD guys - last I checked nobody from M$ had
: > mentioned them in an ftp bench.
:
: Yeah, sure. I realise that. But if resources are going to be put into
: setting up a test bench, I'd hope that it's not too narrowly focussed.
:
: The FreeBSD ftp results prompted Larry to promote the idea of a
: "community" test bench. Let's see if we can expand the terms of
: reference.
:
: Are you suggesting that promoting a more flexible test bench is a bad
: idea?

Nobody is suggesting that. What they are saying is this:

a) the Gbit setup is more realistic in real life. If you needed more
bandwidth, you get a bigger pipe. We could optimize for 10 100Mbit
cards, but that would be pretty silly, right? 99.9% of the machines
in the world can't support that many cards.

b) the point of performance work is **NOT** benchmarks. The point of
performance work is to get better performance. I could care less if
we do well or poorly in benchmarks against anyone. What I do care
about (and you should as well) is that we do well under real workloads.

c) if you ask for too much, you get nothing. Let's stay focussed on the
original setup. If that actually happens, yields results, the kernel
improves, asking for more hardware is easy.

Thanks,

--lm

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/