Re: [RAM] Renumbering impossibility: TSL/SSL certs, DNS delegation etc.

[Robin Whittle <rw at firstpr.com.au>]

Hi Olivier,

By "end-users" I meant those "end-users" who run networks and which
need multihoming, or the ability to choose another ISP without a
renumbering effort which in their view is excessively costly or
likely to cause more disruption than their organisation can handle.

I wasn't referring to people at home with NAT firewalls in ADSL
modems who have never heard of an IP address or a configuration file.


> I think that we strongly need to distinguish between addresses used as
> locators and addresses used as identifiers. Users will probably want to
> keep the same identifier, but they don't care about the locators.

The addresses which matter to these end-users are those of the
packets which are in their network.  How the routing system, LISP
etc. gets them to the network is not something they care about -
unless those methods cause inefficiencies, difficulties with
fragmentation, or break Path MTU Discovery by the hosts which are
sending them packets.

Unfortunately, I think LISP, eFIT-APT and Ivip will cause these
difficulties.  But something like this will probably be built of
those difficulties are not as bad as what would otherwise happen
with the number of advertised prefixes growing without constraint.

 - Robin


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