At the IGF2010 in Vilnius, two folk are floating a trial balloon about separating the allocation function from the registry services function. Currently, these functions are seen as indivisible by the Internet addressing community. In other words, one gets an allocation or assignment from a RIR and the RIR adds the assignment to their database.

The address holder is then responsible for updating the various (inetnum, route and domain) objects in the RIR database that have operational impact.

The question being asked is “Is it time for a split between allocation and services for Internet number resources as was the case for domain name resources?” My answer is no, that comparing IP address distribution to domain name services is an apples to oranges comparison, it wouldn’t be helpful to those of us in the developing world, and would increase cost and complexity to a system that is already an ideal model of cooperation, collaboration and communication in Internet Governance.

First of all the DNS, while a “Critical Internet Resource”, is really just a “layer of misdirection” and one that has spawned a multi-billion dollar industry. Unfortunately that kind of monetisation has created all sorts of unpleasant activity in the DNS, from typo (and domain) squatting to fast flux hosting and other “stupid DNS tricks”. On the other hand, the IP address distribution system is open, transparent, efficient and makes its policies in a bottom up manner that is the epitome of the “Internet Model” as espoused by ISOC, amongst many others in the Internet Community. Subverting that model for profit is anathema to the health and well being of the global Internet IMO.

From a development perspective, which the IGF2010 is discussing at great length, Dr. Milton Mueller at the “Development Agenda Approach to Internet Names and Numbers workshop” suggested that there would be opportunities for entrepreneurs in the developing world to build companies providing “post allocation services”. The reality here in Africa is that while there are hundreds of ICANN accredited registrars, only 4 (four) are African. While I agree with Milton that development is best done via economic opportunity, there is no reason to suspect that African entrepreneurs would establish IP address registry services businesses any more than they have built registrars. I suspect that such businesses will be developed by folk in the USA mostly, as that is where the greatest expectations seem to be in terms of IPv4 address market.

Adding costs to a system that is currently done on a cost recovery basis (by non-profit RIRs) is also a no-brainer in terms of development. Already, our Internet costs in Africa are an order of magnitude greater than those in the EU or USA, I don’t see how adding to the costs of LIRs (ISPs usually) will help in increasing Internet penetration. Currently, Local Internet Registries (LIRs) are expected to maintain WHOIS records for the resources they hold.

These services are WHOIS management (making assignments to customers from of an allocation and recording it in the appropriate database), Internet Routing Registry updates, and the reverse delegation (rDNS) of IP address blocks. While it is currently possible to outsource these activities to others, I would not want to see them being separated from the RIRs (full disclosure: I offer these services to African LIRs as an independent consultant, but prefer to teach them HOWTO do it themselves).

For Allocations and Autonomous System Numbers, these database objects are maintained by (mnt-by in RIR speak) the RIRs themselves to prevent hijacking, so giving third parties access to change these objects would be problematic at best. For rDNS, the RIRs maintain the zone files in the in-addr.arpa domain for those resources under their administration. In other words, the RIRs are responsible for providing the platform so that resource holders can administer their own reverse delegation records. IP address holders have traditionally had the right to administer their own reverse delegation. It seems that involving a third party would just be adding cost and complexity.

I don’t understand what type of policy change (if any) these folk would require. I don’t see a future whereby IANA re-delegates reverse DNS for /8 blocks away from the RIRs. That’s just a non-starter in the current ICANN environment. Since they are at the IGF, which has no policy formulation role, I assume they are just trying to introduce the concept to non-technical folk. I don’t see as how this is going to gain any traction amongst the Internet technical community.