Re: Waste will kill ipv6 too

[Jima <nanog () jima us>]

On 2017-12-21 08:58, Christopher Morrow wrote:
> On Thu, Dec 21, 2017 at 10:16 AM, Jason Iannone <jason.iannone () gmail com>
> wrote:
>
> > M&A plays into this too.  By my calculations, CenturyLink controls at
> > least 17 million /48s.  How many sites does CenturyLink provide
> > service to?  I'm gonna go out on a limb and say it's not 17 million.
>
> there are less than 17m households served by centurylink's residential
> product? really?
> each of those could be considered a site and then get a /48.

Speaking non-hypothetically, there's some shoddy network address 
management at play there. I can state for a fact that there's at least 
one /48 (and I imagine many more) in Jason's list that hasn't been valid 
for over three years. The IPv4 side of that circuit (a /25) is also 
still SWIP'd -- not that it's meaningfully usable in the DFZ.

Therein lies a (minor, I hope) flaw in Job Snijders' proposal to use 
ARIN OriginAS data for determining routing authorization: ISPs have to 
not suck at cleaning up SWIP entries for dormant circuits. I (as a 
customer) tried to get my employer's entries removed three months ago, 
but no one cared enough to follow up.

(Also, I doubt the vast majority of CenturyLink's residential customer 
base a) has non-tunneled IPv6 or b) receives a /48.)

If anyone from AS209 wants to clean up those SWIPs, they're welcome to 
ping me off-list. :-)

- Jima