[rrg] [Fwd: I-D Action:draft-carpenter-renum-needs-work-00.txt]
Flinck, Hannu (NSN - FI/Espoo)
hannu.flinck at nsn.com
Tue Oct 28 08:52:00 PDT 2008
Well, why can't you carry the prefix info and metrics in IGP and use
RAs/ND only at the first hop router (the router that has a host route to
the client)?
- Hannu
>-----Original Message-----
>From: ext Teco Boot [mailto:teco at inf-net.nl]
>Sent: Monday, October 27, 2008 21:41
>To: Flinck, Hannu (NSN - FI/Espoo)
>Cc: rrg at irtf.org
>Subject: RE: [rrg] [Fwd: I-D
>Action:draft-carpenter-renum-needs-work-00.txt]
>
>Hi Hannu,
>
>> However, the pdf
>> presentation leads me to wonder the interactions of the extended RAs
>> with IGPs. They should not be transparent to each other, shouldn't
>> they?
>
>There are some options here. My current thought is use IGP to
>the maximum extent (so I can blame the IGP:-). All packet
>forwarding is using FIB lookup. 1st on DA, if no_match on BR
>address. I'm not sure yet on what a BR should inject in the
>IGP, the full prefix it owns or just its configured prefix on
>an interface. I'll write the options down in the I-D on BRDP
>Based Routing (give me to the end this week).
>The BRDP metric (UPM) is related to metrics of IGP, but not
>one to one. UPM is for both directions and would be "uniform"
>for different IGPs.
>
>> The use of ND RA in the MANET network might be optimal, but not
>> necessarily so in the intra domain without taking into consideration
>> interactions with IGP.
>
>I think it is always a compromise. Using ND is neutral on
>IGPs. And IPv6 routers have ND functions.
>Hosts may use BRIOs, ND is preferred over IGP for this reason.
>
>Teco.
>
>
>
More information about the rrg
mailing list