[arin-ppml] quantitative study of IPv4 address market
mike at nationwideinc.com
Fri Sep 7 16:09:24 EDT 2012
>Would it make you feel better if MS (and DoD for that matter) were to
inject routes for these networks into the DFZ that went either to a
firewall with a "deny ip any any" policy or to null0 on a router with
no internal connection to that organization's network?
>I'm not suggesting gratuitous abuse of DFZ slots, only illustrating
(using the converse of the illustrations that JC and Owen have used)
that a routing table entry in the public Internet and value as
globally unique addresses as something that you can not see from your
particular vantage point... have nothing to do with each other.
Yes, I am keenly aware that ARIN does not require allocations to be routed.
I was pointing out the oddity of a nearly 99% unrouted ratio in combination
with the other oddities of the transaction as potential evidence of an
invalid justification test.
Sure there are edge cases, but a simple glance at the ratio of advertised to
allocated address for the total pool of allocations indicates that
allocations are as a whole routed at 90% since 1997, per the somewhat dated
I'm sure the intricate MS network was able to be demonstrated to ARIN in
such a way that ARIN decided that a mix of 16 /24s, 4 /23s, a /22, 2 /21s, 4
/20s, a /18, a /17, and 9 /16s was exactly what was required for the MS
internal non-connected infrastructure needs which could not be carved out of
And that demonstration happened in the matter of time between ARIN becoming
aware of the deal and the 30 days after that the Motion to Approve was being
heard by the judge.
And that turned out to be the exact distribution of address space already in
place in the negotiated deal between MS and Nortel that was being sent to
the judge for approval.
Removing the needs test for transfers and having ARIN simply book legal
transfers would eliminate the need for any more fishy justifications in the
More information about the ARIN-PPML