[ppml] Policy Proposal 2005-1: Provider-independent IPv6

[Jeroen Massar]

[large snip of a very good writeup]

On Thu, 2006-04-27 at 01:03 -0700, Owen DeLong wrote:
[..]
> 	building a scalable routing system or waiting for the network
> 	to implode, I would hope that they will start working towards
> 	a more scalable solution, such as ID/LOC splits.

Indeed. A possible solution like shim6 might take still quite a number
of years to be developped and deployed, but the current solution
"injecting those PI spaces into the global routing table" can scale most
likely for the time being.


One thing that especially large ISP/Transits's (I saw something like
Sprint/MCI/Verizon 'voting against PI) seem to be confusing about is
that there is a difference between "address space" and "routing
entries". "PI" does not mean that it will end up in the DFZ, it will
most likely, but it might not. A company can decide to internally use
the PI space, so that they have globally unique address space, but when
accessing services on the internet they might proxy (or NAT or shim6)
this address space and thus effectively connect from the address space
their upstream provider provids them. For the coming years though indeed
most PI blocks will pop up in the DFZ. But what is the issue there? If
you are a large transit, you simply let your client PAY for announcing
that prefix into the DFZ. No pay? -> No route. That is if you really are
that scared that your hardware can't handle it, and if it can't,
something I said to a certain "Large Tier 1 Transit" before: UPGRADE!
(and when the economics don't work out for you, you might want to start
rethinking if you are in the wrong business or not ;)

Greets,
 Jeroen

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 313 bytes
Desc: This is a digitally signed message part
URL: <https://lists.arin.net/pipermail/arin-ppml/attachments/20060427/3e26f377/attachment.sig>