[arin-ppml] IPv6 Allocation Planning

Owen DeLong owen at delong.com
Mon Aug 9 20:05:59 EDT 2010

>>> One problem, of course, is that ISPs are used to planning address
>>> consumption on 6 and 12 month scales, not decades. They have no
>>> practical experience to guide them with longer range planning.
>> While this is true, it's a relatively minor problem of education.
> Let me put it to you this way: Looking forward ten years, at the IPv6
> experience, knowledge and understanding you will have then, what
> percentage of it do you already have today?
In all honesty, probably somewhere between 40 and 60%.

> And for IPv4?
Probably close to 100%.

> I'd bet many of us are past the 80% mark for IPv4. There are
> refinements to learn and probably a few surprises, but we're already
> highly knowledgeable.

> If you claimed more than a single-digit percentage for IPv6 then you
> either overestimate your current abilities or underestimate what
> you'll yet learn. The events that define the shape, understanding and
> use of IPv6 are still more in our future than in our past.
I think you actually underestimate my capabilities and IPv6 experience.
I suppose I can consider this part of your retort for pointing out the
inconsistency in your own statements.

> Any policy we write now will reflect that inexperience. And will
> necessarily change as we gain greater skill.
Of course... Policies are not intended to be permanent or immutable.

I don't see how this makes a case against teaching providers how to
make better projections for longer-periods of usage of IP address space.

>> To prevent this from impacting the routing system, yes, providers
>> should be discouraged from disaggregating this space. I believe
>> that the community is, generally, quite capable of doing this through
>> education.
> Because "education" has been particularly effective at suppressing
> disaggregation in IPv4?
Of course it hasn't, and, I pointed out many of the reasons why. Mostly
they have to do with scarcity and the mentality of scarcity. Scarcity
makes aggregation impossible. Scarcity doesn't apply to IPv6 unless
we artificially apply it through misguided mechanisms like slow start.

> As things stand, our policies partially usurp ISPs' ability to create
> and enforce disaggregation policies amongst themselves. They've asked
> us repeatedly to seek policy which at a technological level empowers
> them to set routing policies independent of our address allocation
> practices. In many ways it isn't even about how many routes there are
> in the table, it's about giving the ISPs the power over routing policy
> instead of keeping some of it for ourselves.
ISPs can set any routing policy they choose as things stand today.
The problem is that their customers would be unhappy and they can't
coordinate the changes without crossing the line of the Sherman Anti-trust
act (or their local equivalent), so, they're in a catch-22. That's not about
how ARIN policy is impacting them, that's about the reality of the legal
framework in which they operate.

> We can do much better than we're doing now and the sooner we start the
> less of a legacy of our error they'll have to carry in their TCAMs and
> TRIEs.
Hence my desire to produce a policy that maximizes the potential for
one ISP one prefix.

You're making very good arguments in favor of my proposal while
claiming to oppose it. I am very confused by this.


More information about the ARIN-PPML mailing list