T-Mobile’s oddly coincidental IPv4 address acquisition

As followers of our blog know, there is an intense and important struggle going on under the surface of IP address allocations. ARIN, the regional Internet registry for North America, is conflicting with companies who would like to broker the market for legacy IPv4 address blocks.

Most legacy blocks are in North America. ARIN has no real contractual authority over them, because they were allocated before it existed and most of the legacy holders haven’t signed any contracts with it. Yet those legacy allocations constitute about 40% of the total IPv4 address space; if they are traded outside of ARIN’s more restrictive transfer policy regime, based only on what buyers and sellers agree, ARIN could lose a great deal of its power to govern the (IPv4) address space. For ARIN, this could be a matter of survival – they could turn into the ITU of IP addressing (I can’t resist saying that!). The stakes are just as high for the buyers, sellers and brokers of IPv4 address space. Large amounts of money and better access to address assets are in play.

What’s in danger of getting lost in this struggle is the uniformity, transparency and accountability of IPv4 address allocations. ARIN now has an incentive to deviate from its policies and individually negotiate needs assessments and contracts in order to prevent the defection of address buyers. We saw this tendency in the Microsoft-Nortel deal. ARIN rushed into the bankruptcy court just as the deal was being concluded to offer MSFT a special, individually negotiated “Legacy RSA,” with stronger rights over the address assets, when ARIN’s transfer policy clearly states that any buyer of addresses is supposed to get a standard RSA contract.

Now there’s another suspicious episode. In mid-February, address broker Addrex announced that it had an entire /8 for sale. Addrex wanted to sell this block completely outside of ARIN’s transfer policy for what would probably be hundreds of millions of dollars. ARIN on the other hand has made it clear that it wants to prevent any transfers outside of its own policy regime, and that it will use its leverage as the authoritative IP address registry to refuse to update the Whois records of any such transfers. This refusal to update the Whois is intended to threaten the routability of any acquired addresses, which might scare off buyers – but it could also damage the integrity of the IP Address registration database as well. It’s a rather unpleasant game of chicken.

T-Mobile was one of the firms potentially interested in buying a /8. Two weeks after the /8 was announced for sale, on February 27, ARIN commenced a sequence of seven allocations to T-Mobile, spaced roughly two weeks apart. The allocations are shown in the table below.

Allocation History

Of course, we can’t prove that ARIN made some kind of a sweetheart deal with T-Mobile to lure them away from Addrex. But what does seem clear is this:

1. ARIN allocations act as a substitute for purchased legacy blocks, and vice-versa.
2. If you are an organization that wants more IPv4 addresses, but don’t meet ARIN’s needs assessment criteria or some other policy constraint,  you will be tempted to acquire addresses from the legacy market.
3. ARIN has a strong organizational self-interest in finding a way to keep you away from the legacy market and inside its own RSA regime. It means fee revenue and it means continued authority.
4. ARIN has made it clear that it will use its organizational resources (e.g., the Whois database) to actively discourage market transfers outside of its own policy framework.
5. ARIN’s applications of its needs assessment criteria is secret; no one outside ARIN can see what information was exchanged to establish whether an organization needs more addresses or not, and if so, how many. No one can see what negotiations took place between ARIN staff/board and a major company that wants addresses.

Together, those five propositions create a situation where strategic bargaining between ARIN and major IPv4 address seekers is likely to take place. If ARIN has addresses available in an environment of intensifying scarcity, and is competing for authority with an open market, it would make sense for it to discriminate in its allocations in favor of entities that might defect to the legacy market, rather than applying its needs assessment criteria and allocation policies in a neutral, nondiscriminatory manner. Again, we lack the information to prove that this happened in this case – but we also lack the information to be confident that it did not.