OpenBSD Journal

BSDCon Europe 2002 Paper available

Contributed by jose on from the from-europe-with-love dept.

Rob Sessink writes:
"With the presentation given by Philipp Buhler and Henning Brauer on performance tuning of OpenBSD network servers at BSDCon Europe, they have made a the paper available. "

This looks like a pretty cool paper, I've been interested in tuning the performance of my OpenBSD systems for a while now. Thanks for the paper, guys, and the link.

(Comments are closed)

  1. By Anonymous Coward () on

    i posted this in the ebsd message before!

    i dont know what my point is :)

  2. By Dom De Vitto () on

    It's got some great tips, and these guys obviously know the kernel inside out. Cute idea for IRQ sharing too, counter intuitive, but I bet it makes a really BIG difference with quality cards.

    It's nice to see other folks kernel configs and sysctl settings too, makes me warn and fuzzy to see others use similar settings.


  3. By Anonymous Coward () on

    quite nice. thanks!

  4. By RC () on

    The section on which NICs perform best was rather brief. I'd like to hear everyone's preferences.

    I mostly stick with Linksys LNE100's, as they are very cheap (~$15), are compatible with practically everything (DEC/tulip compatible), perform quite well, and it's incredibly rare that I come across a defective card.

    Anyone else?

  5. By Anonymous Coward () on

    I'd love to see FreeBSD's device polling ported to OpenBSD. If you want to see why this is needed, do a ping -f to an OpenBSD box on 100baseT.

  6. By Anonymous Coward () on

    application. Many of pf's newer features were needed, but it was not possible to wait for 3.1- Release, as the previous OpenBSD 2.9 firewall running IPFilter had saturated the processor at near 100% utilization at peak usage times, and delays were being noticed.

    They get to talk about OpenBSD performance, and take a jab at IPF, at the same time... Yay1

  7. By Anonymous Coward () on

    Darren Reed correctly pointed out that is actually a paper on network performance monitoring than really tuning.

    pb@ seems to agree. He says they lack time to provide a more detailled paper.

  8. By Anonymous Coward () on

    Experimental zero-copy for TCP and UDP transmit-side (using page-loaning) would be nice as well.

    And of course pipes with page-loaning (this is on art's todo list).


Copyright © - Daniel Hartmeier. All rights reserved. Articles and comments are copyright their respective authors, submission implies license to publish on this web site. Contents of the archive prior to as well as images and HTML templates were copied from the fabulous original with Jose's and Jim's kind permission. This journal runs as CGI with httpd(8) on OpenBSD, the source code is BSD licensed. undeadly \Un*dead"ly\, a. Not subject to death; immortal. [Obs.]