[arin-ppml] A compromise on legacy space?

[John Santos]

On Wed, 10 Sep 2008, Owen DeLong wrote:

> 
> On Sep 10, 2008, at 8:10 AM, John Santos wrote:
> 
> >
> > This argument totally ignores what happens if ARIN allocates a
> > "slot in the database" to someone when the number corresponding
> > to that slot was allocated by a predecessor to a legacy holder.
> >
> > You can't make the issue of legacy assignments go away by
> > sophistry.  It still has to be dealt with.
> >
> 
> First, I don't think that Stephen's argument is sophistry under
> the definition of sophistry:
> 	The use of fallacious arguments, esp. with the intention
> 	of deceiving.
> 

Actually, it wasn't Steven's argument that I was responding to,
but Kevin's summary.  Unfortunately, it came as an attachment
rather than in the main text, and when I replied, it got lost.

I'll see if I can retrieve it...

[later...]

Kevin said:

> When it comes down to it nobody will force anybody to do anything.  People
> without any agreement or who do not keep registration data current can
> continue to use whatever numbers they want, they just won't be in the
> database.  ARIN cannot force Legacy users to enter into an agreement, and
> the Legacy users cannot force ARIN to list them in the database.
> 
> Networks (the majority of the Internet community) who choose to use this
> database will be able to easily communicate with others who use this
> database.  If someone wants to communicate with a network that is not 
> in the
> database their administrator is free to make a manual route table 
> entry.  So
> long as everyone between the two endpoints has the same entry it will route
> just fine.
> 
> The ARIN database is not a magic thing, it is not a law, it is not even a
> requirement.  It is just a tool we use to make our lives easier so we don't
> have to type route entries in to our routers or maintain our own database.
> ARIN is just a service to save us all work. It is useful because many
> networks choose to use it.  You could route just fine without ARIN if you
> could just exchange route entries with everyone you need to communicate 
> with and all their transit peers.
> 
> The reason this functions and works is cooperative anarchy. This is a good
> thing. Nobody forces anyone to do anything and if you cooperate and play by
> negotiated guidelines (RFC's) then everything works.  Simple, cheap and
> effective.  What could be better?

The gist of it seems to be that by thinking of it as
"slots in a database" instead of thinking of it as an allocation
or as an IP network number, ARIN could rewrite the rules to do
anything it wants (i.e. ignore legacy assignments.)  I think that
it makes no difference at all what way you think of it.  This is
what I refered to as sophistry.

Kevin, if you were just attempt to rephrase the argument using
the terminology of the "slots in a database" model instead of
the "allocations or assignments" model, then I apologize for
misunderstanding.

I still disagree with removing the legacy assignments (under most
circumstances) and think ARIN would encounter extreme difficulties
(i.e. law suits and/or government intervention) if it attempted
to do so, though.


> What happens if ARIN allocates a "slot in the database" to
> someone else when the number corresponding to that
> slot was allocated by a predecessor to a legacy holder
> is, indeed, the crux of the matter, but, also, happens to
> not truly be under ARIN's control.  The possibilities
> as I see them are:
> 
> 1.	A critical mass of ISPs believe the new ARIN entry
> 	and the legacy holder is SOL.
> 
> 2.	A critical mass of ISPs believe the legacy holder
> 	and the new registrant in the ARIN database is
> 	basically SOL.  Likely, ARIN is also discredited to
> 	some extent, and things break.
> 
> 3.	A nearly equal mix of ISPs choose each side and
> 	nobody is happy.  All sides lose credibility and
> 	the government decides they've let the amateurs
> 	control this internet thingy long enough.
> 

4. Both the legacy holder and the new registrant (and possibly
all legacy holders as a class action) sue ARIN for "conspiracy
in restraint of trade"...  (I was going to put a smiley here,
but I think *no one* on this list would be happy about that.)

> I don't believe anyone will actually support the idea of
> simply reusing legacy space which was previously
> assigned just because the legacy holder does not have
> a contract with ARIN in the immediate future.

That seems to be the opinion of most, but not all, of
the people here.  There are *some* (maybe a very small
minority, maybe not) who seem to want to revoke *all*
legacy space now if not yesterday, some even if it's
under an LRSA.

> 
> I do believe that as ARIN outreach to legacy holders
> becomes more successful and more legacy holders
> do sign agreements with ARIN, the remaining legacy
> holders will face increasing pressure and decreasing
> credibility.
> 

Probably true, which is one reason I would really like to get
something I can live with.

> I'm not sure that's a good thing, but, I think it is a likely
> scenario.  I think it is in both ARIN and the legacy
> holders' interests to get as many of those resources
> protected in the hands of the legitimate legacy holders
> through a contractual relationship with ARIN as soon
> as possible.

I do think we are getting closer.  Unlike some other legacy
holders (apparently), I don't even mind the provisions that
I could lose my "slots in the database" if I sign the LRSA
and then stop paying and don't respond to email or snail
mail or phone calls or any other attempt by ARIN to contact
me.  It seems that if I terminate for cause (or even without
cause if that variation becomes policy and part of the LRSA),
I have to tell ARIN that I'm doing so.  If I've signed the
LRSA and then I just vanish, stop paying and become uncontactable,
it seems reasonable for ARIN to (eventually) assume the space
is abandoned and reclaim it.


As another carrot to signing, it looks like I wouldn't qualify
for PI IPv6 space under current rules (too small), but I think
I could get it if I were an LRSA signer.  My provider seems not
to be doing anything with IPv6 yet, at least I can find *no*
mention of it on their web site, so ISP-provided IPv6 is not
currently an option.  I want to start testing with IPv6 and one
of my customers, who is *also* an ISP (a very big one but not
*my* ISP; I don't know if I can divulge their identity so that's
why I'm being vague) with whom we have a private T1 line carrying
IPv4, is actively investigating IPv6 for its internal network
and when they mandate a transition, our private line will probably
also have to be converted or at least the routers at both ends
dual-stacked.

This is both a carrot for me (I get my IPv6 space) and for all
the people out there urging faster IPv6 adoption.  (Assuming
I haven't completely misunderstood the IPv6 terms.)


> Owen


-- 
John Santos
Evans Griffiths & Hart, Inc.
781-861-0670 ext 539