What can the Root Zone take?

[Rubber Duck]

>>KIEREN McCARTHY: And second question from Don Hollander, is there a technical limitation on the number of names that can be put into the root?
>>CHRIS DISSPAIN: Is there a technical limitation on the number of names. Ah, okay.
Go ahead, Vint.
>>VINT CERF: This is a frequently asked question and often highly debated question. I think the answer is the root can stand substantial expansion, technically. You can probably put 10,000 names in the root and have it not collapse.
The real issue is not so much the physical ability to put the domain names into the root zone, but, rather, all of the work that's associated with having something in the root zone. The work that's involved in making changes to the root zone entries for movement from one I.P. address to another, and any other interactions that might be necessary between ICANN and the entity that's to whom the top-level domain is delegated.
So it's the administrative processes of both getting through an application and also in just day-to-day operation and servicing of a TLD operator that might limit the total that can be sustained.
But the technical limitations are not nearly, I think, as constraining as the procedural ones.

http://losangeles2007.icann.org/file...t1-29OCT07.txt

[jacksonm]

For sure the root zone can be substantially expanded. The work the root DNS servers do today is simply processing the same small set of queries over and over - queries for ccTLD and gTLD domain (extensions) NS records. Yes, a top level extension is nothing more than a domain itself which has been delegated to someone else.

The root servers are behind load balancers, e.g. A.ROOT-SERVERS.NET is not a single server, but a farm of servers load balanced behind a single IP address. Every new extension which is added to the root causes X tens of million new queries per day to the root server network. Not a major problem, IMO - they can add more and more backbone root servers.

.

[Rubber Duck]

I think the key point is that there is a lot of work associated with managing strings directly within the Root and extrapolating to very large numbers via aliasing is likely to be easier to manage via DNAMES, so it would seem that DNAMES is still live as a medium term solution, but there is no desperate hurry to go to the next step as there are no significant immediate advantages from doing so.