New ARIN ipv6 allocation policies

David Barak thegameiam at
Sun Sep 3 15:22:25 CEST 2006

--- Pekka Savola <pekkas at> wrote:

> On Sat, 2 Sep 2006, David Barak wrote:
> > --- David Conrad <david.conrad at> wrote:
> >> As we've learned from IPv4, I fear swamps are
> >> forever...
> >
> > But so what if they are?  ... At that point,
> > you're at what, 35K unique routes?   What
> equipment
> > will have trouble with this?
> Two things:
>   1) Because of "everyone else has it, we must also
> be able to get it" 
> mentality.  Else they'll threaten with suing based
> on restraint on 
> trade and other ludicurous reasons.  A precedent on
> creating swamps 
> sets pressure to continue to make the swamp even
> "swampier" (e.g., by 
> relaxing allocation rules even more, not even trying
> to get the act 
> together, etc.)

All of those incentives exist today.  The net effect
of those incentives is a ~200/month growth in ASNs. 
Regarding lawsuits, a simple, clear policy is the
single best defense against lawsuits; how could
anything be simpler than "PI allocations are made at
/48 boundaries to organizations which already possess
and qualify for a unique RIR-issued ASN"?

I agree that efforts to make the swamp swampier need
to be resisted, but I don't think the pressure to
swampify will be all that horrible.
>   2) 35K unique sites could result in a lot of
> dynamicity in the 
> routing tables.  

Not more than any of the existing 180,000 unique
routes in the IPv4 DFZ, eh?  What we're basically
doing with this new IPv4 policy is returning to
classful routing: a /32 is for ISPs, and a /48 is for
multihomed enterprises.  Nice and easy, and if
necessary, ISPs can segregate the swamp away from the
rest of their routers.  

I wonder if you have seen BGP
> Update report? [1] 
> Multiply 35K by, for example, 500 or 1,000 updates
> per day, and you've 
> got quite a bit of churn at your hands -- churn that
> you can do 
> nothing about because folks at DFZ believe they have
> the right to do 
> anything they want with their advertisements.
> [1]

One thing I notice in that report is the complete
absence of enterprises - all of the names listed are
themselves service providers.  How much swamp space is
causing all of that?  More likely is some of the
intentional deaggregation (consider the case of AS
6197, who advertise twice as many routes as they need,
and are also near the top of the updaters list)...

-David Barak

David Barak
Need Geek Rock?  Try The Franchise:

Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 

More information about the ipv6-ops mailing list