Waste will kill ipv6 too

Excuse the top post, but this seems to be an
argument between people who understand big
numbers and those who don't.

Nobody needs to worry. I promise to reserve the last /32 out of my /29
assignment. When the world has run out of addresses, I will start to sell
would consider a /64 to be equal a /32 IPv4 address. This would make a /56
assignment equal to a /24 IPv4 minimum assignment.

Historically we spent about 3 decades before running out of IPv4 space. So
my scheme should be good enough for some additional decades of IPv6.

I just hope nobody else does the same. That would be bad for my business



And if a medical breakthrough happens within the next 30 years? Nanobots
that process insulin for the diabetic, or take care of cancer, or repair
your cells so you don't age, or whatever, perhaps the inventor things ipv6
is a good idea for such an endeavour. a nanobot is microns wide, and there
will be billions per person, hopefully not all on the same broadcast
domain.In fact, as you saay, we should treat /64s as a /32 and a /64 for
ptp. So each nanobot gets a /64. 10B nanobots per person times 20B people =
oh, crap, we've exhausted the entirety of ipv6 an order of magnitude ago.
Let alone the fact that actual usable ipv6 /64s is 2 orders of magnitude
below that.

(the time has finally arrived)

Obligatory xkcd ref: https://xkcd.com/865/

Just how many nanobots can dance on the head of a pin?
  - Brian

Think in terms of system architectures where the address space is fully
consumed when empty to more than 20 decimal places. Because we're idiots
and actually designed it that way.

From: John Lightfoot <jlightfoot@gmail.com>

Excuse the top post, but this seems to be an
argument between people who understand big
numbers and those who don't.

No, not exactly. It's also about those that
think in current/past network terms and those
who are saying we don't know what the future
holds, so we should be careful.

which means 79 octillion people...no one
alive will be around

Stop thinking in terms of people. Think in
terms of huge numbers of 'things' in the
ocean, in the atmosphere, in space, zillions
of 'things' on and around everyone's bodies
and homes and myriad other 'things' we can't
even imagine right now.

Sure, but likely zillions of ocean sensors will share a few /64s rather than getting a /48 each.

Do you really think each person needs more than a thousand or so subnets for their wearable sensors? If not, then 1 of the many /48s they can safely consume has them covered.

Can I see a possible future in which homes actually need /48s? Sure. But we’ve got more than enough /48s to do that.

As I’ve said many times before, let’s see how it goes with the first /3 doing things as designed and intended. If it turns out to consume that 1/8th of the address space while I’m still alive, I’ll happily help build more restrictive allocation policies for the remaining virgin 5/8ths and the fractions of the 1/4 of the address that have a very small number of special use carve-outs (0::/3 and e000::/3).

Given that we still have more than 500 /12s free in the first /3 20 years into the process, I’m thinking we aren’t likely to have that issue.


Giving each nanobot a pair of /64s would be absurd. Maybe they aren’t all on the same link (there are no broadcast domains in IPv6), but likely a few /64s would cover each person.