IPNG Directorate (NGDIR)

Reported by Scott Bradner/Harvard University

The IPng Directorate held an open meeting on Monday afternoon.  The
session was a joint meeting with the IPNGWG BOF. The format was that the
IPng Directorate sat at the front of the room and invited questions from
the attendees:


   o What is the expected/necessary address assignment efficiency of
     IPng addressing proposals?  This question should be asked tomorrow
     at the Address Autoconfiguration BOF (ADDRCONF).

   o What is the relationship of private IPv4 addresses (``net 10'') and
     IPng addresses?  This needs to be covered.  Large users need to get
     large blocks of addresses with no questions asked.  There is also
     concern about the relationship of mobility and autoconfiguration.
     What is the effect of mobility and autoconfiguration of addresses
     with authentication (how do you authenticate with a changing IPng
     address unless you use an EID?).

   o The opinion that 16 byte addresses are too big was expressed.

   o What is the trade-off between time (getting the protocol done
     quickly) versus getting autoconfiguration and security into the
     protocol?  Autoconfiguration and security are important carrots to
     get people to use IPng.  The trade-off between making IPng better
     than IP (so people will use it) versus keeping IPv4 to be as good
     as it can be.

   o We sound like we are not quite done, sort of like IP TOS. True, but
     we are older and wiser now.

   o DNS is not very robust, let's not depend upon it (it is ``already
     breaking'').  The network is growing, adding autoregistration, and
     security depends upon DNS. Much of the problem with DNS is the
     implementation, not the specification.

   o Concern was expressed about IPng and scaling, e.g., exponential
     address space growth, address space density, and flows.  The claim
     is that in practice the limit in routers is destinations per
     second.  There is also concern about source routing---what service
     provider will allow users control over routes?  EIDs are evil.

   o It was stated that in order to put flow state into routers we need
     to be able to aggregate flows.

   o Concern was expressed about address autoconfiguration.  Host
     configuration is a host management issue.  Management software at a
     high level is needed.

   o The opinion was expressed that we must focus on/stick to the
     schedule.  Commit to keeping to the schedule and solving problems
     that we know how to solve.


During the IPNGWG BOF part of the meeting, Steve Deering presented the
working group charter.