Le 07/12/2011 02:15, Lorenzo Colitti a écrit :
On Tue, Dec 6, 2011 at 12:34, Alexandru Petrescu
<alexandru.petre...@gmail.com <mailto:alexandru.petre...@gmail.com>>
wrote:
I think 4191 is desirable if the host is a Host on cellular, not a
Router on cellular (tethering IPv6). A pure Router does not consider
receiving RAs other than for logging purposes, thus ignoring the
potentially present RIO.
IPv6 CE routers already listen to RAs and use them for routing
information (see RFC 6204). There's no reason a phone can't do the
same.
Yes, it sounds logic. But, looking again at RFC6204 - that is a set of
requirements. It does not say how to implement it.
Moreover, it does not request the use of RFC4191 on the WAN interface of
CE (or I can not find it requiring it so).
For PD, it does not tell what to do with the delegated prefix (it should
say adding routes in CEA and in WAN).
For reasons like this I think RFC6204 may be enough for CE but not for
other settings like a telephone.
Additionally, RIO (Route Information Option) does not seem to encode
the address of next-hop (it seems left to determine elsewhere, not
clear its relationship with the prefix in RIO).
The next-hop is the router that sent you the RA. This is good because
it implies fate-sharing - if the router goes away or is not capable
of carrying your packets any more, then the RA goes away as well.
If you want to send traffic to a specific prefix through another
router on the link, then that router must send the RA itself. It
doesn't have to announce itself as a default router if it doesn't
want to be - it can just carry traffic to that prefix.
Fate-sharing may be an advantage for CE, but can also be seen as the bad
single point of failure (RA and routing functionality tied together) (if
RA software on a Router dumps core, its routing kernel still works ok,
thus another router could inform about this route by sending RA about it.)
Alex
_______________________________________________
mif mailing list
mif@ietf.org
https://www.ietf.org/mailman/listinfo/mif
