On 03/11/15 01:21 +0000, Lizhenbin wrote:
> Hi Folks,
> 
> I proposed my concern on the implementation of interoperability between LDP 
> and SR again in IETF. I also would like to remind you of my draft again:
> 
> https://tools.ietf.org/html/draft-li-spring-compare-sr-ldp-rsvpte-00
> 

With the hopes of clarifying Lizhenbin's comments which I found a bit
hard to follow both at the mic and in this email I went through this
draft and found the most applicable section to be 4.2 (proxy egress) in
the draft above versus Section 6 which he referenced.  I believe this is
the concern he is raising (Lizhenbin - please feel free to correct me).

In inter-as option C and like technologies, there are two ways to
distribute received label information from ASBR/PE to other PE's, either
via BGP label-unicast or IGP (make other networks PE's look similar to
this network's PE from label stack standpoint by redistribute prefixes
into IGP and have PE advertise via LDP which results in a 2 label
approach versus 3 label in BGP case).  I'll refer to these as C1 (IGP)
and C2 (BGP).

draft-ietf-spring-segment-routing-ldp-interop-00 already does a good job
of describing that option C/CsC/Seamless MPLS can be supported in it's
Section 6 (using C2), but it could add a bit more text I think admitting
that it can't support using a C1 model as many providers have deployed
and therefore an extra label will be incurred.  At least I can't think
of an easy way to support having multiple ASBR's handle mapping these
dynamically learned prefixes from the other AS to the same SID...

-Jon

_______________________________________________
spring mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/spring

Reply via email to