[RAM] Re: I-D ACTION:draft-lear-lisp-nerd-01.txt

[Eliot Lear <lear at cisco.com>]

Brian E Carpenter wrote:
> Hi Eliot,
>
> I guess the RAM list is the place for this. I really only
> have two comments at the moment.
>
> > 3.1.  NERD Record Format
>
> I don't think it's OK only to document the IPv4 format.
> By the time any of this becomes reality we'll be in real need of
> IPv6 support. It's much better to design it in now, and include
> it in the calculations.

Hmm... You'll note the AFI in the record format.  The analysis is IPv4, 
and it is worth mentioning how IPv6 would impact it.  Your analysis 
below seems accurate.

> I think it "only" doubles your EID and RLOC sizes - 64 bits should
> do fine; even a /64 mask seems unlikely in practice.
>
> > 8.  Deployment Issues
> >
> >    While LISP and NERD are intended as experiments at this point, it is
> >    already obvious one must give serious consideration to circular
> >    dependencies with regard to the protocols used and the elements
> >    within them.
>
> Indeed one must. I strongly suggest getting rid of your dependency
> on DNS for NERD bootstrapping and update. I see no way to assert
> with certainty that http://www.example.com/eiddb/ can be resolved without
> any indirect dependency on LISP

I would argue that name servers and NERD web servers involved should 
have globally routed addresses.  This resolves the interdependencies.  I 
see no reason, by the way, why everyone must use LISP ETRs and PI 
space.  If you are single homed, what's the benefit?

Eliot

_______________________________________________
RAM mailing list
RAM at iab.org
https://www1.ietf.org/mailman/listinfo/ram