[ppml] Policy 2002-5
>On the same vein, how do we feel about this rewording? Also, I again
>that 12 months is way too long a time, especially if the organization is
>already not using a part of the space.
I think that we should set aside this issue until it can be more clearly
formulated and discussed. One issue that I see is that this is trying to
do two different things and I believe that we need two more focused
policies to accomplish that. Firstly, we want to contact all legacy users
of IPv4 space and get them to become part of the system, i.e. keep in
touch and register their up to date contact information. Secondly, we want
to organize the pre-ARIN IPv4 space in a more logical manner because we
can probably achieve a higher usage rate by doing so. And maybe thirdly,
we want legacy users to pay for service. At a minimum there is the
in-addr.arpa service, but a case can be made for part of the routability
attribute being provided by being legitimately registered with ARIN.
Therefore, let's drop the specific proposal 2002-5 and let's begin to
discuss how we can better address its contents in a more focused and
I would suggest that there be a program to contact all holders of legacy
IPv4 space and that we should tell them that they MUST register their
contact information with their RIR or else relinquish their allocation.
This is not a hardship for any address space holder. Some organizations
may claim that their use of the space is outside of the public Internet
and therefore they should be left alone, however I disagree. They should
still at minimum register their contact info and the status of the space.
Whether or not they should legitimately continue to be given exclusive
usage rights to the space is something that we should not be dealing with
at all right now. We just need to know who has the usage rights to every
single fragment of the IPv4 space.
If, in fact, we are going to reorganize the layout of the legacy
allocations such as the swamp, then I believe that we need more discussion
and some joint action with the other RIRs. Any policy relating to this
should be identically worded with all RIRs. I believe that it is a good
idea to swap space only if the result is a less chaotic map of the IPv4
space at the level of RIR allocations to members. What I mean is that
today we know that in certain ranges, the RIR allocated all blocks at /19
or larger. In other ranges the RIRs allocated at /20 or larger. If we can
reorganize the swamp so that there are defined ranges in which all
allocations are /21 or larger, /22 or larger, /23 or larger and /24 or
larger then this would be a good thing. But we won't really know how big
to make these ranges until we have an idea of who has legacy usage rights
and still wishes to retain those rights. Also, once we decide to
reorganize the space through swapping, we should have at least a rough
plan for offering new allocations with the same sizes. And perhaps we
should even be issuing smaller blocks than /24 in a defined range for
small multihomed networks.
And then there is payment for service which is an issue that strikes at
the heart of what a registry is and what it does. Here, I believe we need
a vision that goes beyond matters such as swamp cleaning in the IPv4 space
but also encompasses the future when IPv6 is so widely deployed that we
have decided to abandon the IPv4 space entirely. People will still be
using IPv4 for lots of things but we won't have any need for a global
registry for IPv4 anymore.
At that time, I still see some value in having RIRs and I see that there
are services for which fees should be paid. ARIN will still be the
maintainer of the single authoritative database that identifies the
organization which has legitimate rights of use for any specific IP
address. For various reasons, I feel that ARIN should be flexing its
muscle a little more in this area. One thing that I believe ARIN should do
is to provide a new form of routing registry which identifies who has the
usage rights to every block and which references any local routing
registry which that rights holder may be operating. It should be possible
for every network operator to verify their incoming route announcements by
querying such a registry either in realtime or by mirroring the database.
Whether or not anyone does in fact do such validation is not important nor
is it important whether or not they trust ARIN well enough to hook the
registry directly to their routers. The important issue here is that ARIN
should make an authoritative database available for queries and mirroring
and that ARIN should enforce this database by removing the data when an
organization is no longer in good standing.
In any case, it would be nice if you all would change the subject lines of
any replies to this message because there are really 4 separate threads
that could be started from these suggestions. Thanks.
-- Michael Dillon