[ppml] IPv4 "Up For Grabs" proposal

[Ted Mittelstaedt]

>-----Original Message-----
>From: ppml-bounces at arin.net [mailto:ppml-bounces at arin.net]On Behalf Of
>James Hess
>Sent: Thursday, July 05, 2007 5:55 AM
>To: ARIN Address Policy
>Subject: Re: [ppml] IPv4 "Up For Grabs" proposal
>
>
>> 4) "Up for Grabs" IP space will be usable by any organization needing
>> IPv4 numbering.  None of the RIR's will provide any sort of mediation
>> between competing organizations wanting to use the same IPv4 space,
>> except for that provided for in #2
>
>It's not that ARIN can't do that, it's that I don't think it makes
>sense to do that.
>
>Essentially, I wonder how "no stewardship at all," by intending to
>incite chaos
>over some addresses, counts as responsible stewardship, as per
>ARIN's mission.
>

Saying you aren't going to steward something doesen't mean your advocating
chaos.

>
>However, in case the policy were implemented that way...
>
>Once random orgs start trying to pick at "up for grabs space", I worry
>if there's
>much to keep the up-for-grabbers away from doing the same, trying to
>use space that was actually legitimately assigned by the current
>registry, but
>the org doesn't "think" is being used.
>

The same situation exists today.  What prevents it is that most responsible
networks don't allow end-node AS's to announce prefixes that are not listed
in a registries database, and presumably responsible networks don't
originate such advertisements.

It is also important to keep in mind that one of these days IPv4 will
be declared obsolete, and the RIR's will stop keeping track of it, at
that point it will effectively be "up for grabs"

>I expect ARIN could lose legitimacy not only with the legacy holders,
>but a lot of
>people out there, who rely on there not being total and utter
>addressing chaos.
>
>
>Recall.. ARIN/etc is not the actual mechanism that allows or disallows
>an organization
>from using address space.
>

Yes, it effectively is, because so many transit AS's rely on it to
determine whether or not an advertisement is legitimate or not.

>The registry itself can't prevent two determined parties from trying
>to use the same
>addresses, that is not the function of ARIN.; Only if they are both
>registrants, can
>it help, and that is only done only by making sure not to assign the
>two registrants the
>same addresses.
>


If 2 entities try to use the same space, and neither is listed in
a record as being the owner, then both are going to have problems and
no network is going to help them solve them (by erecting blocks or
whatever) until one of the entites signs an RSA with an RIR and becomes
the assigner of record for that space.

Please reread my original post.  I said that only legacy space where
the legacy owner refused to sign an RSA for it would be "up for grabs"

If the legacy owner does sign a RSA then the space is now allocated to
them, and is no longer up for grabs.  If another party tries using it
then they will be frozen out by the network operators who can easily
see that the space is allocated to the legacy holder now.

>
>Possibly, if providers found that ARIN said "anything goes" for the
>legacy assignments,
>the providers would just come up with their own ad-hoc rules to pick
>up where ARIN left
>a big hole.
>

Quite possible.  Of course, the providers would certainly require the
legacy holders to sign an RSA with the new entity that controls the
legacy assignments, as well as pay for maintaining the records, so I
don't see why any legacy holder would prefer to do it this way.  Either
they fund an RIR alternative or they fund the RIR, either way, the
free ride is over.

>I.E. some of the legacy registrants would become further solidified,
>when their providers
>develop ad-hoc filters to discard attempts by "rogue orgs" to announce
>prefixes that
>would be "up for grabs" according to ARIN
>(rogue orgs being anything other than the legacy holder).
>
>Otherwise, the IP address would no longer be globally unique, making
>it useless to everyone.    Now all the legacy holders would suddenly
>have a justifiable need for new addressing,  they may even be able to
>justify larger assignments than they had before.
>
>Exhaustion could occur even more quickly at that point.
>

OK let's be a little more specific, please.  Your saying a legacy holder
might abandon a legacy IPv4 assignment and sign an RSA for a new
unsullied IPv4 assignment with a RIR?  Why would they do this?  The
cost would be the same as if they just went ahead and signed an RSA
for the legacy addresses they already are using, and they wouldn't
have to renumber.

>
>ARIN only provides the service of assuring registrants that their
>assignments are
>unique among other registrants, which is a pre-requisite for their networks
>being able to communicate with each other.
>
>It is providers themselves that respect whatever ARIN's registry
>says a range
>is assigned to.
>
>I think they know better than to accept "up for grabs".
>
>Either that means the addresses become useless, or the consequence
>is the legacy
>holder gets them permanently, even if the legacy holder later
>decides to return
>addresses, filters may remain in place all over the world.
>
>It's not hard to have chaos, but how can you have it one little range
>of addresses
>and really be assured of not have it  all over the place?
>

We already have such a situation, as it's been already documented that
spammers are using legacy IPv4 that is NOT advertised, advertising
it themselves, and spamming from it.

Ted