[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: RFC 5006 status
On Wed, 17 Mar 2010 17:13:27 +0100
Michael Cardell Widerkrantz <mc@hack.org> wrote:
> Fred Baker <fred@cisco.com>, 2010-03-17 16:18 (+0100):
>
> > This is a structured question for the community.
> >
> > Jari Arkko tells us that he is getting requests from various sources
> > to take RFC 5006 to Proposed Standard. It is now experimental.
> >
> > http://www.ietf.org/rfc/rfc5006.txt
> > 5006 IPv6 Router Advertisement Option for DNS Configuration. J. Jeong,
> > Ed., S. Park, L. Beloeil, S. Madanapalli. September 2007. (Format:
> > TXT=26136 bytes) (Status: EXPERIMENTAL)
> >
> > (2) Vendors, please advise on implementations. Are there any? Has
> > interoperability been demonstrated?
>
> Implementations of RFC 5006 known to me:
>
> - Server side: radvd,
>
> http://www.litech.org/radvd/
>
> - Client side:
>
> My own radns, slowly getting there,
> http://hack.org/mc/hacks/radns/
>
> rdnssd, now maintained as a part of ndisc6:
> http://rdnssd.linkfanel.net
> http://www.remlab.net/ndisc6/
>
> radvd works well with both radns and rdnssd.
>
radvd and radns have also worked successfully for me.
> > (3) Operators, enterprise and/or service provider, please advise on
> > deployment experience.
>
> I have so far only used this with a manual configuration in radvd, but I
> think this will typically be used in a home or SOHO environment where
> the ISP gives addresses to DNS resolvers to a home/small office router
> with DHCPv6 and that router, in turn, sends this information to end
> nodes with the RDNSS option in RA.
>
I've been thinking a bit about the RDNSS verses stateless DHCPv6
question a bit in the last few weeks.
I think there is value in the RDNSS option, however I think it is only
really useful for low end embedded devices - ones that are resource
constrained to the point where running a stateless DHCPv6 client might
be too much overhead, although that is a very low point though these
days. Sensor networks could be an example.
(I'm not sure if my perspective is correct, however I've started to
think that the original boundary between RAs and Autoconf was a
"bootstrap" connectivity boundary. RAs provided enough information for
IPv6 to get to the point of being able to send packets beyond the
local router, Autoconf was then to fill in other more user oriented /
application supporting parameters. DNS certainly does fit into the
latter category, although I can see it also being considered a bootstrap
connectivity parameter, which is why the RDNSS option was created)
However, I think the other parameters that RDNSS don't support, such as
SIP servers or NTP servers, are going to be "widely" useful. I think
it's inevitable that all telephones in people's homes will eventually
be IP phones, and they'll need SIP server and possibly time server
addresses. It's possible that all clocks on people's homes will
eventually be "IP clocks" so announcing NTP servers may also be a wide
spread requirement.
In conjunction with the likely requirement of DHCPv6-PD on CPE, I think
stateless DHCPv6 clients will be nearly ubiquitous on IPv6 devices.
> --
> http://hack.org/mc/
> Use plain text e-mail, please. OpenPGP welcome, 0xE4C92FA5.
>