Last Call: 'Connecting IPv6 Islands over IPv4 MPLS using IPv6 Provider Edge Routers (6PE)' to Prop

Francois,

[clipped...]
 
> I wonder if the current text of rfc2858bis offers
sufficient  
> flexibility to allow this. Quoting:
> "
>   Address Family Identifier:
>       This field in combination with the Subsequent
Address Family
>       Identifier field identifies the Network Layer
protocol associated
>       with the Network Address of Next Hop and the
semantics of
>       the Network Layer Reachability Information that
follows.
> "
> 
> The above text suggests to me that a given AFI/SAFI
pair dictates a  
> single protocol for the next hop, while we are talking
here about  
> potentially allowing different next hop protocols which
are  
> disambiguated via other means (such as next hop
length).

We are talking about the existing IETF standards track
documents
(l2vpn, rt-constrain) that for a given (AFI, SAFI)
combination allow 
the next hop to be either IPv4 or IPv6, and use the next hop
length 
to disambiguate between the two.

> Am I mis-interpreting rfc2858bis text?
> Is it conceivable/wise to tweak this text into
something slightly  
> more flexible along the lines of:
> "
>   Address Family Identifier:
>       This field in combination with the Subsequent
Address Family
>       Identifier field identifies the _set of possible_
Network Layer  
>       protocol associated
>       with the Network Address of Next Hop and the
semantics of
>       the Network Layer Reachability Information that
follows.
> "

I personally have no problems with your proposal, except
that
the same tweaking should also be applied to the text on
SAFI.

Yakov.

_______________________________________________
Softwires mailing list
Softwiresietf.org
http
s://www1.ietf.org/mailman/listinfo/softwires