[ previous ] [ next ] [ threads ]
 
 From:  Kris Maglione <bsdaemon at comcast dot net>
 To:  Chris Buechler <cbuechler at gmail dot com>, m0n0wall dash dev at lists dot m0n0 dot ch
 Subject:  Re: [m0n0wall-dev] Re: [m0n0wall] The future
 Date:  Thu, 13 Oct 2005 12:22:22 -0400
Chris Buechler wrote:

>The -dev list is intended for that purpose, but it gets
>a lot of noise.  pfSense has a private committers-only dev list that
>isn't archived anywhere, and is invite-only, and has turned out to be
>a great resource for the developers (you can be a lot more open and
>not watch what you're doing so much when your words won't be forever
>archived by Google).
>
I definately agree with the "won't be archived by google" part. I do
think that there should be at least one public -dev list so the main
list doesn't get clogged by comments and conversations meant for the
developers by people without access to the private list.

>On FreeBSD, for those who haven't followed and messed with 4.x vs. 5.x
>vs. 6.x as much as I have:
>FreeBSD 4 is a hard act to follow, but there's no question we can't
>continue with it.  4 has one of the fastest general-purpose TCP/IP
>stacks ever written.
>
Do you have any input on Dragonfly? I haven't done any testing or seen
any benchmarks, is doing some good work with the networking stack.
Within a few months, they claim that they will have eliminated Giant and
made the entire stack threaded... I'm not sure whether this will affect
single processor performance in a good or bad way, but it should be
great for SMP. At the very least, it merits the same level of testing
that 5.3 got, though I'm not sure that I like the idea of the beta
flip-floping between BSDs between releases.

Just for reference, here's Dragonfly's status page (i.e. list of
improvements):
http://www.dragonflybsd.org/status/diary.cgi

Also, from the latest release notes:
"The greatest progress has been made in the network subsystem. The TCP
stack is now almost fully threaded (and will likely be the first
subsystem we remove the BGL from in coming months). The TCP stack now
fully supports the SACK protocol and a large number of bug and
performance fixes have gone in, especially in regard to GigE performance
over LANs."

>So Dru and Colin would recommend, this was in July, that if
>deploying a server in August you skip the current "stable"
>"production" 5.x and use a beta of a non-stable release on a
>production server?!?  Yes, shows great confidence in 5.x there.
>  
>
Or it says "Sure 5.x is great, but 6.x is f***ing awesome" or some such
thing. :)

Anyway, 6.0 is nearing release (rc1 is out), and should be stable by the
time that m0n0wall 1.3 is released, so it's worth consideration.