[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: OpenBSD PPPoE performance test results


Thanks for your feedback.

On Mon, 2 Dec 2002, Dom De Vitto wrote:
> Unfortunately, can you, in retrospect see the problem with your OpenBSD
> figures:
> Linux 2.2 ... 100%  pppoe 95%, nc 5%
> Linux 2.2 ... 50%  pppoe 40%, nc 10%
> Linux 2.4 ... 26%  system 24%, nc 11%
> NetBSD 1.6 ... 23%  system 15%, nc 17%
> NetBSD 1.6 ... 23%  system 15%, nc 17%
> FreeBSD 4.7 ... 100%  ppp 80%, nc 15%, interrupt 5%
> OpenBSD 3.2 ... 99%  system 87%, pppoe 65%, ppp 23%, nc 7%

You can clearly see that OpenBSD is pushing its limits here. It simply
could not transfer the data any faster. Therefore a simple "CPU load 100%"
would have been enough, but I wanted to be more specific. Nevertheless,
these values are taken from the output of top.

More realistic and accurate results could have been obtained by performing
the test on 100Mbit/s network with three computers (the P90 acting as a
router between the PPPoE server and "the LAN"). Testing the maximum
transfer speed alone would have provided more accurate results and netcat
wouldn't have had to be ran on the PPPoE client itself. However, at the
moment I lack proper resources for such a test - I don't have enough
quality 100Mbit/s NICs I could use in the test.

Still, I'm not subscribed to the list, but I thought replying to the list
would be the best practice.

Juuso Raitala

Visit your host, monkey.org