[governance] ICANN Board Vote Signals Era of Censorship in Domain Names

[Karl Auerbach]

Vittorio Bertola wrote:

>  However, on resources that 
> are not infinite (and sure, TLDs now are artificially scarce, but would 
> not be infinite anyway)

To make that somewhat quantitative:

Peter Deutsch (formerly of Bunyip and Cisco) and I ran an experiment a few 
years ago, on what today would be considered pretty wimpy computers.

We took the .com zone that we had (I can't remember how we got it, or when, but 
it had a lot of of names in it.)

We elevated those names so that they were each a TLD.

We loaded it into bind and measured performance.  We had to add memory to the 
machine but once we had enough, it loaded and response was amazingly good.

Then I wrote a program to generate synthetic root zone files so that I could 
create root zones of any size and with a mix of character-string lengths (just 
to make sure I didn't accidentally get an advantage from some hashing mechanism 
somewhere).

And then I had another program that generated queries, both hits on names that 
were in the zone file and misses on names that were not.  I could adjust the 
hit/miss ratio.  We measured responsivity and query loss.  I don't think we did 
any really heavy traffic loads because we made the assumption that UDP based 
DNS queries could be spanned across multiple servers using standard 
load-balancing front-ends.

We got into the millions and millions of TLDs, but never found an upper bound.

We did not follow our scientific training - we didn't keep good lab notes and 
our observations were more subjective than objective - so to do it right we'd 
need to locate and resurrect the pieces and do it again.

So take the result with a grain of salt, but it does appear that we can readily 
have tens upon tens of millions of TLDs without the server software melting.

The limit on TLDs is probably more based on the chance of administrative error 
rather than on a hard technical limit.  But .com teaches us that it is possible 
to administer a rather large (60,000,000 name zone) without a lot of 
administrative errors.

If we assume a rather low number, well within the range of both the technology, 
the software, and administrative processes: 1,000,000 TLDs.

If we aim at that number, we could allocate 10,000 per year and we would hit 
that highly conservative limit of today in year 2107, a century from now.

So I think it is useful to drop the conceit that we need to conserve TLDs until 
the numbers reach at least 4 orders of magnitude greater than the 200..300 TLDs 
that exist today.

However, since even a billion TLDs would be less than the number of people who 
might want one, we do need some mechanism to allocate.  And that would bring us 
to the auction/lottery discussions we had a while back and also to some very 
good academic analysis of auctions and lottery methods of allocating TLDs.

		--karl--

____________________________________________________________
You received this message as a subscriber on the list:
     governance at lists.cpsr.org
To be removed from the list, send any message to:
     governance-unsubscribe at lists.cpsr.org

For all list information and functions, see:
     http://lists.cpsr.org/lists/info/governance