ietf
[Top] [All Lists]

Central registries (was RE: Chinese IPv9)

2004-07-19 10:28:50
Paul, this IPv9 hoopla strikes me as another research project harping
the nationalistic chord in order to get funding. This is not exactly
news. It was a common undertone in many European research proposals in
the 1980's and 1990's, and it is also a classic line in NSF or DARPA
proposals. The officials in the Chinese government may fall for that
line a few times, but I believe that they are smart and will eventually
allocate their grants based on technical merit rather than non-technical
arguments.

There is however an interesting technical point behind all these
discussions of number allocations. The general Internet architecture is
largely decentralized, but we have accepted to rely on a few centralized
functions. The obvious ones are DNS names and IP addresses, but there
are many others, such as port numbers and generally the various
registries held by IANA.

Centralized registries are expedient, and are not a big concern when the
network is small, or when the central authority is virtuous. However,
the network is big and the central authority becomes a locus of power.
The history text books teach us that loci of power attract politicians
and politician-friendly profiteers, and the Internet does not appear to
be an exception.

It seems that we, the IETF community, have been complacent to
centralization and have dug ourselves in a centralization hole. We may
hope to get out of it by ensuring that ICANN remains in charge and
remains virtuous, but that goes very much against all historic
precedents.

When in a hole, one should obviously first stop digging: that would mean
a moratorium on the creation of IANA registries. One should also think
hard about technical alternatives to central registries. In some case,
that may mean a slightly larger field in a protocol format, so a large
random number can be used instead of a short registered number. In other
cases, like name resolution, that may require a technical break-through.
But we should definitely think about it!

-- Christian Huitema

_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf


<Prev in Thread] Current Thread [Next in Thread>