[ipv6hackers] the end is near (or for IPv6: the beginning)

[Owen DeLong]

> On Jan 29, 2014, at 1:22 PM, Michael Richardson <mcr+ietf at sandelman.ca> wrote:
> 
> 
> Owen DeLong <owend at he.net> wrote:
>>> Up to now, whenever I see an IPv6 rollout planned in a NATed IPv4
>>> environment, this usually means adding IPv6 with a stateful firewall to
>>> the existing IPv4 with NAT (which will be kept).
> 
>> Sure, but soon we'll also start seeing greenfield deployments of IPv6
>> without IPv4 of any sort.
> 
> Doubtful.
> 

Inevitable, actually. When there is no more IPv4, if you want to deploy something, you'll have to use IPv6.

> Too hard to get IPv6-only allocations from ARIN.

Not true.

> Not cheaper.  Not easier in paperwork.

A: Depends... If you are an ISP, then it's included in the same price. If you are an end-user, then IPv6-only is cheaper than IPv4+IPv6 now.

B: Not true at all. IPv6 paperwork is very easy. For an end user, it boils down to:

	1.	I am multihomed.
	2.	I have X sites.
	3.	Compute n for 2^n=X, and round n up to the nearest multiple of 4.
		If 2^n-X < 0.25X, then n+=4.
	4.	Request a /(48-n)

	For an ISP, it boils down to:

	1.	Compute the number of aggregation points at your principal aggregation level for
		your networks hierarchy (POPs, etc.). Round up to a nibble boundary as above,
		with 25% minfree. Consider the required number of bits to be X.
	2.	Compute the number of end-sites served by your largest aggregation point from (1)
		above. Round up to a nibble boundary as above, with 25% minfree. Consider the
		required number of bits to be Y.
	3.	Submit request for /(32-(X+Y)).

Owen