[arin-ppml] efficient utilization != needs basis
I had an epiphany the other day, I believe we are frequently equating
"efficient utilization" with "operational need" and "needs basis" in our
policy discussions. In todays world of IPv4 that is understandable, and
mostly true in IPv4 policy. I've seen numbers claiming there are 5
billion devices connected to the Internet today. So it is obvious that
with only 4 billion addresses in IPv4, "operational need" has a strong
correlation with "efficient utilization" for IPv4.
However, back in the day of class based IPv4 "operational need" and
"efficient utilization" were not strongly correlated, other issues like
DFZ size were far bigger issues. But there has always been a "needs
basis" for assignments.
In IPv6 "efficient utilization", at least on an address by address
basis, wasn't even a design consideration, in fact one reason for 128
bits was to make this a non-issue for a long time. Since it is not even
a design consideration it is not a necessary as a primary attribute for
IPv6 number resource policy. So everyone needs to disassociate
"efficient utilization" from there definition of "needs basis" at least
as related to IPv6. But I believe there still should be a "needs basis"
for address allocation in IPv6 it will just be defined in a completely
Note that HD-Ratio is by definition a measure of address assignment
efficiency, not necessarily a measure of need in a IPv6 world where
"operational need" and "efficient utilization" have little or no
With all that said I think we need to figure out what a "needs basis"
really means for IPv6. For devices connected to the Internet, AKA the
Web, or public broadband service, in a IPv6 world the primary challenge
I see is DFZ growth. I believe IPv6 addresses intended for this purpose
for the immediate future will need to be based on a hierarchical
routing, with mostly provider aggregated addressing, and limited
provider independent end-user addressing, probably based on multi-homing
However, there are other IP technologies coming that may or may not be
connected to the Internet and will possibly form other internets. These
internets will probably have other challenges and possibly form other
address hierarchies independent of the Internet as we know it today.
But, they will almost assuredly need globally unique addresses too.
Some examples are; Smart Grids, emergency response networks, and one
that has been functioning from more that a decade is the Automotive
Network Exchange (ANX).
Why do these need globally unique addressing? As was discussed on PPML
already, Smart Grids have a enormous number of potential devices that
will have addressing needs, just by that sheer size globally unique
addresses are the only really practical solution. Emergency response
networks will need globally unique addressing, because the consequences
of address conflicts will likely be measured in lives lost. Globally
unique addresses really make thing like the ANX much easier, and there
are many other private inter-AS networks that exist today and I only see
more and more in the future.
These all have different challenges and I bring them up in this context
because they all also have "operational need" and as we to define a
"needs basis" for IPv6, we must find a definition that encompasses the
"operational needs" of these other internets too.
David Farmer Email:farmer at umn.edu
Networking & Telecommunication Services
Office of Information Technology
University of Minnesota
2218 University Ave SE Phone: 612-626-0815
Minneapolis, MN 55414-3029 Cell: 612-812-9952