> 
> -----Original Message-----
> From: Soliman Hesham [mailto:[EMAIL PROTECTED]
> Sent: mercredi 17 mars 2004 09:38
> To: Pascal Thubert (pthubert); Jari Arkko
> Cc: [EMAIL PROTECTED]
> Subject: RE: Multiple DRs on a link
>
>
>  > => There are many different reasons. I sent a verly long
>  > email about this to nemo (monet back then). One simple
>  > scenario is that you might be walking around with a PAN
>  > that happens to have 2 MRs on a single link (e.g. a laptop
>  > and a mobile phone). The two MRs could share the same ingress
>  > link and have different egress links. For instance your laptop
>  > might have a WLAN card and your mobile might have a cellular
>  > interface. Once you walk into an airport lounge or starbucks
>  > you'll suddenly have a multihomed PAN. By definition, each MR
>  > must have a separate home prefix. So each will advertise
>  > a different prefix on the ingress side.
>  >
>  > Hi Hesham:
>  >
>  > I fail to understand the "by definition" here. Nemo does not
>  > prevent 2
>  > different MRs from registering the same MNP. It's like
>  > multiple parallel
>  > routes. What's not duplicated is the Home address... In
>  > fact, I proposed
>  > a test to check that both registrations actually end up in a
>  > same link,
>  > otherwise there's a problem equivalent to the DAD problem in
>  > MIP6. Not
>  > much success on that proposal, though
>
> => Hmm, that's interesting. I must have missed that in the
> spec. So, my concern is similar to yours, how does the HA
> check whether the MRs are in the same place?? Seems like
> a dangerous feature to have. I'm surprised this wasn't discussed
> during LC.
>
Well, Nemo requires that some authorization be run, that's not described
in the spec. Such a test could take place there, or the authorization
could bar a second registration for a same prefix, I guess.

Pascal


--------------------------------------------------------------------
IETF IPv6 working group mailing list
[EMAIL PROTECTED]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to