[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Overloading RA [Re: RFC 5006 status]



Seiichi Kawamura åé:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I fully agree with Brian here.
I would choose DHCPv6 inside my ISP/data center
network to do any dynamic node configurations where necessary.
Traceability is a MUST.
But I would like to have SLAAC in other areas specifically
services that connect nodes
that don't have robust amount of resources.

So keeping SLAAC simple, but with the
options that are absolutely necessary to
get connected(default router and dns)
without relying on other protocols or manual
configuration, would be the best thing.

I second.

xing

Regards,
Seiichi

Brian E Carpenter wrote:
On 2010-03-19 11:25, Ralph Droms wrote:
You still miss my point.  I don't think sending config info in RAs is
necessarily a bad idea.  The IETF has worked from an architectural
guideline of providing one way to carry config.  Some network operators
would rather use one protocol to carry config, be that ND or DHCP.  It's
just a different philosophy.

I do think the idea of carrying config info in RAs is being sold with
FUD.  I also think the ways in which adding more config info to RAs
ought to be thought through fairly carefully, based on some of the
IETF's relevant experience with DHCP.
What I notice is that DHCP has become very heavily loaded with options.
Although one can deploy DHCP without using all those options, it has
become a complex thing. SLAAC at least has the advantage that it's
stayed simple; in fact the only overloading so far is RFC 5006. I think
we'd need a very convincing argument to depart from that simplicity,
as opposed to simply saying: if you need to convey arbitrary parameters
to hosts, use DHCPv6, which is intended for that purpose.

A DNS server address is not an arbitrary parameter; along with a default
router, it's the second address you *must* know in every host. So adding
it to SLAAC seems consistent. Adding arbitrary parameters to SLAAC would
be a whole other approach, and I'm not sure I understand the motivation.

    Brian
- Ralph

On Mar 18, 2010, at 1:10 PM 3/18/10, Gert Doering wrote:

Hi,

On Thu, Mar 18, 2010 at 12:40:15PM -0700, Ralph Droms wrote:
I don't  happen to share your opinion that "[t]he IETF has solidly
messed up this part of IPv6 by delaying things for 10 years or so,"
No need to share any opinion here.  Just try to run an IPv6-only network
with Windows, Linux and MacOS clients, and consider whether the IETF
might have given vendors better guidance.  Or finished one or the other
RFC 5 years earlier.

As an operational person, this endless bickering between the RA camp and
the "what do you need RA for that, when DHCP can do this all along" has
FAIL stamped all over it.

Gert Doering
       -- NetMaster
--
Total number of prefixes smaller than registry allocations:  150584

SpaceNet AG                        Vorstand: Sebastian v. Bomhard
Joseph-Dollinger-Bogen 14          Aufsichtsratsvors.: A.
Grundner-Culemann
D-80807 Muenchen                   HRB: 136055 (AG Muenchen)
Tel: +49 (89) 32356-444            USt-IdNr.: DE813185279
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (MingW32)

iEYEARECAAYFAkui1sEACgkQcrhTYfxyMkLcFgCfYrw8qJBRg1RusLOHtAt6OOfj
uH4An0deFXQJJ0X1a/2UJ7ux0a9vkbAn
=W/oE
-----END PGP SIGNATURE-----