Re: [narten(_at_)us(_dot_)ibm(_dot_)com: PI addressing in IPv6 advances

On Fri, 2006-04-14 at 14:17 -0400, Noel Chiappa wrote:
>     > From: Kevin Loch <kloch(_at_)hotnic(_dot_)net>
>
>     > Nobody in that room would have supported a policy they actually
>     > believed would blow up the Internet.
>
> Who was in the room, BTW? How many of those 60 were from ISP's?
>
> Also, does that group have any commitments from ISP's (particularly the
> large global backbones) to carry these PI addresses? (I assume the group is
> expecting that PI addresses will be supported by the routing, not by some
> as-yet-undefined other mechanism.)
I guess actually that the larger ISP's are not too happy about "PI
space" in general, as with the Aggregate system they can nicely lock-in
customers.

As for "global backbones", check GRH (http://www.sixxs.net/tools/grh/)
and look at the amount of /48's and up seen in the routing tables and
which transits are providing these in BGP today. Most, if not all, do
this thus this will pose no problem whatsoever.

Let's see how it all will take of, when GRH shows a large increase in
new allocations we'll notice quickly if this is a success or not.

We'll max out the 16-bit ASN space first anyhow ;)


In the very very very long run, your undefined mech, we might end up
using these PI /48's for 'identifiers', using the upstream's /48 as the
locator space. No renumbering will thus be required, only some border
changes, and the tricky bit: a signaling protocol for notifying the
other end who we are so they can map it back to the PI /48. IHMO that
will become a real solution for many problems. Announcing multiple
prefixes and using source-address selection for instance is far from
pretty.

Greets,
 Jeroen

signature.asc
Description: This is a digitally signed message part

_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf