Last Resort Registries

  > poole at eunet.ch writes:
  > > 
  > > Perhaps you don't have big companies in Switzerland, whatever do I know
  > > about Switzerland. But there are cases when Big Company connect to a
  > > small outside company, providing something special and as Small Company
  > > don't connect to Internet they must use RFC 1597 and eventually
  > > Small-2 has been told the same story and use the same first numbers
  > > from RFC 1597, but when Big Company want to have a link to Small-2
  > > as well, this becomes a problem.
  > 
  > If you'd followed the IETF list you would know that I made a comment
  > pointing these problems out a -long- time ago, in particular from the
  > point of collisions RFC-1597 is substantially worse than picking
  > addresses at random (and no better than the time honored tradition
  > of using addresses from Sun).

There is a serious omission in RFC1597: We did not recommend that people
who forsee external connection requirements should choose their RFC1597
addresses at random. We just assumed that people would be clever enough to
figure that out by themselves. This will be fixed in the next version of 
the RFC.  However we did say:

   Groups of organisations which foresee a big need for mutual
   communication can consider forming an enterprise by designing a
   common addressing plan supported by the necessary organisational
   arrangements like a registry.

If they choose to have Bjorn do that for them it is perfectly OK.
However Bjorn should be *very* clear about the fact that this address
space is still "private" and that there is no guarantee of uniqueness
outside the set of organisations participating in this
voluntary scheme! Of course the guarantee of uniqueness from public 
address space remains.

Can we close this side discussion?

Daniel