Mailing List Archive

Efficiency of IPv6 NUD
Has anyone got any practical results on how fast NUD occurs for
<linux/windows/blah> systems when they have multiple valid default nexthops
learned through RAs and one (or all but one) of those nexthop routers then
goes offline without cleanly sending a withdraw.

Any relevant radvd timing tweaks would also be of interest.
Re: Efficiency of IPv6 NUD [ In reply to ]
Le 25/05/2012 10:41 AM, Oliver a écrit :
> Has anyone got any practical results on how fast NUD occurs for
> <linux/windows/blah> systems when they have multiple valid default nexthops
> learned through RAs and one (or all but one) of those nexthop routers then
> goes offline without cleanly sending a withdraw.
>
> Any relevant radvd timing tweaks would also be of interest.

See here:
http://packetlife.net/blog/2011/apr/18/ipv6-neighbor-discovery-high-availability/

NUD is no VRRP.

Simon
Re: Re: Efficiency of IPv6 NUD [ In reply to ]
On Friday 25 May 2012 11:06:32 Simon Perreault wrote:
> Le 25/05/2012 10:41 AM, Oliver a écrit :
> > Has anyone got any practical results on how fast NUD occurs for
> > <linux/windows/blah> systems when they have multiple valid default
> > nexthops learned through RAs and one (or all but one) of those
> > nexthop routers then goes offline without cleanly sending a
> > withdraw.
> >
> > Any relevant radvd timing tweaks would also be of interest.
>
> See here:
> http://packetlife.net/blog/2011/apr/18/ipv6-neighbor-discovery-high-avail
> ability/
>
> NUD is no VRRP.
>
> Simon

Thanks.

Absolutely agree that NUD is no substitute for a real HA protocol - but in
situations where a few seconds of unreachability is acceptable in a
failover event, this seems pretty decent.

My one concern is replicating this with radvd since if I remember rightly,
it mangles the times you pass in the config according to a formula.

Regards,
Oliver
Re: Efficiency of IPv6 NUD [ In reply to ]
Applications like this are generally open source so you can contribute packages, fixes and enhancements upstream.

Jared Mauch

On May 25, 2012, at 2:02 PM, Oliver <olipro@8.c.9.b.0.7.4.0.1.0.0.2.ip6.arpa> wrote:

> My one concern is replicating this with radvd since if I remember rightly,
> it mangles the times you pass in the config according to a formula.