Subject: Re: IPv6 Fast Forward
To: None <tech-net@netbsd.org, tech-kern@netbsd.org,>
From: Liam J. Foy <liamfoy@sepulcrum.org>
List: current-users
Date: 02/28/2007 21:26:41
On 28 Feb 2007, at 20:35, Liam J. Foy wrote:

> Hello folks,
>
> The following patch integrates IPv6 Fast Forward into the
> NetBSD kernel. It should apply to current just fine - if not
> it'll be some minor nit which you can just directly report
> to me and I'll roll a new patch.
>
> ftp://ftp.netbsd.org/pub/NetBSD/misc/liamjfoy/ip6_flow.c.diff
>
> You can control it via a new sysctl:
>
> 	sysctl -w net.inet6.ip6.forwarding=1 <- of course
> 	sysctl -w net.inet6.ip6.maxflows=256 <- new sysctl
>
> (ip6.maxflows has a default of 256). Also, remember to build
> your kernel with GATEWAY. ip6.maxflows of course controls how
> many flows can be created.
>
> Basically:
>
> IPv6 Fast Forward increases forwarding speed (especially on low
> powered CPUs in embedded devices) by hashing/caching the
> flow (a flow being from a src to a dst). You are basically limited by
> the network card/driver. This is available in IPv4 and IPv6.
>
> Thanks to matt@, joerg@ and christos@ for initial review (as always).
>
> Usual paypal donations for a poor 19 year old student go to
> liamfoy@sepulcrum.org
>
> ;-)

Forgot to say -  It'll be committed in roughly one week time if
no issues are found :-)

		---
		Liam J. Foy
		<liamjfoy@netbsd.org>