[ppml] Restrictions on transferor deaggregation in 2008-2: IPv4 Transfer Policy Proposal

[William Herrin]

On Tue, Mar 11, 2008 at 1:53 PM, Scott Leibrand <sleibrand at internap.com> wrote:
>  Do you think that the IPv4 Transfer Policy should restrict deaggregation
>  of transferred netblocks?  Why or why not?

Scott,

Yes, the policy should restrict deaggregation.

Why? Allow me to spin a hypothetical scenario:

ScottNet, an ISP serving the eastern seaboard, goes bankrupt and the
pieces get auctioned off by the court. The customer base is sold by
locality to a dozen distinct ISPs who agree to honor some portion of
the remaining contracts in order to gain the customers.

ScottNet held 3 /12's. In its infinite wisdom, it allocated /24's to
its individual POPs as needed and then assigned customer prefixes out
of those /24's.

Without a restriction on deaggregation, the /24's assigned to each POP
are transferred to the purchasing ISP for that POP. Those ISPs
aggregate where possible but mostly have to announce the individual
/24's.

The net impact of ScottNet's dissolution is that as many as 12,000
prefixes are added to the DFZ RIB and FIB at a systemic cost that
could exceed $74M/year (http://bill.herrin.us/network/bgpcost.html).

I think we can all agree that would be an unfortunate outcome.


On Tue, Mar 11, 2008 at 2:05 PM, Bill Darte <BillD at cait.wustl.edu> wrote:
> And while you are answering these questions, or just browsing, please
>  state plainly whether you are 'in favor' of this policy proposal, or
>  'opposed'.

Opposed at this point, but I do agree that an address market is
inevitable and ARIN should act to regulate it intelligently. You're
asking good questions and working towards a policy I could support.


Regards,
Bill Herrin




-- 
William D. Herrin herrin at dirtside.com bill at herrin.us
3005 Crane Dr. Web: <http://bill.herrin.us/>
Falls Church, VA 22042-3004