Moving to IPV6

[Robert E. Seastrom]

Mike O'Dell <mo at ccr.org> writes:

> IPv6 addresses are really only 64 bits long,
> no matter what they tell you in the funny papers,
> and even all 64 of those aren't up for grabs arbitrarily.
> this statement is based on the fact that the ability
> to distinguish individual "things" is limited by the
> E164 address in the "bottom" 64-bit half of an IPv6 address.

A correction and then a monograph on IPv6 space.  This is an EUI-64
address, not E.164 address.  EUI-64 is a unique identifier that is 64
bits long.  EUI-48 (its 16-bit-shorter cousin) is used for Ethernet
MAC addresses.  E.164 is a phone number.  See wikipedia articles on both.

I'm a bit at odds with Mike on the analogy here.  You can't make a
direct comparison between IPv6 and IPv4's address space.  About the
closest you can come is to compare a /64 (the normal subnet size) with
a /28 or a /24 (common and traditional subnet sizes) with the caveat
that (subject to neighbor discovery protocol scaling and broadcast
domain size) each subnet can contain an arbitrarily large number of
hosts (about 2^64) rather than 13 or 253 as in the example above.

How big is a /64?  Glad you asked; I love Fermi questions.  If one
were to drain all five Great Lakes and fill them with almond (not
peanut or plain; this is important) M&Ms, you'd have right about a /64
of M&Ms there.  Probably ought to drain and fill the Great Salt Lake
with insulin while you're at it.

Anyway, I digress.  Let's accept Mike's "it's only a /64" as
postulated since we're arguing a range of 2^8 (i.e. 2^64 to 2^72
comparing to a single /32 in IPv4 land - or if you like, a /64 is
equivalent to a /32, a /24, or something in between), which at the
scale we're talking about isn't really that significant.

It took from 1983 to 2011 to use roughly seven eighths of the 32 bit
address space (the rest is unusable for various reasons, and yes I'm
aware that there was experimental use before 1 Jan 83).  Let's call it
30 years, since if we'd had the justification requirements of today in
the early days we'd have easily made it another 2 years.

A straight projection of the same run rate would have us running out
in (30 * (2^32)) or 128849018880 years.  It would also be "wrong" -
the growth in use rate is not linear.  I'm going to go out on a limb
here and say that our sun and our planet, let along the human race,
and particularly let alone IPv6, is not gonna last a hundred billion
years.

A fairly pessimistic projection that includes a lot of slop for waste
would have the number of IPv6 /64s starting out at 64k times the size
of the IPv4 address space (assumption is a /32 allocated for every
classic ASN) and doubling every 30 years.  That gives us runout in 480
years.  Humans will last that long probably, if we don't do anything
stupid.  My guess though is that the utility of the IPv6 address space
will have been overtaken by events by then.  Not much of a market for
quill pens or commercial sized wooden sailing ships these days.

Another way to look at sustainability is to consider that the default
IPv6 end site allocation is a /48, despite the efforts of some people
who think that 65000 subnets allocation for a home network is
excessive (it is, but that's the whole point - you get more than you
would ever need) and have pushed for a default size of a /56 in order
to conserve space.  Never mind the fact that if you handed every human
on the planet a /48, by 2020 (accounting for population growth) you
would have used about 1/30000th of the address space.

Anyway, the *real* problem that IPv6 fails to fix is one that I'm
surprised that Mike didn't mention since he had an IETF draft or three
that would have fixed it.  That problem is one of routing table
growth.  In the middle of the Internet are large routers - not like
the one that you have on your cablemodem, but ranging in size from 2
rack units tall and weighing some 40 pounds to several adjacent 42u
racks and weighing a couple of tons.  They carry the routes for
Internet provider allocations.  This is called the default-free zone
or DFZ, because you don't carry a default route or gateway - you're
supposed to know how to get everywhere.  A full routing table in IPv4
land is 375k routes, approximately.  In IPv6 land it is 4500 routes.
Both are growing exponentially.  Processing power on the routers'
management systems to sort out the routing information is not.  You
can see where this is headed...  eventually...

Only the first eighth of IPv6 space is in play.  There are two paths
you can go (route?) by but in the long run there's still time to
change the road you're on.  I think a famous rock musician once said that.

-r