any interesting/useful resources available to IPv6 only?

Lee Howard lee.howard at
Tue May 7 11:53:44 UTC 2019

On 5/6/19 12:45 PM, Brian J. Murrell wrote:
> But the came I am making is to PHBs, not engineers and I am trying to
> find a path of least resistance.

IPv6 is, on average, 20ms faster than IPv4. I don't know why, I just 
know that the evidence is diverse and compelling that it's true.

A faster web site means people find it earlier in Google search, stay on 
it longer, and buy more stuff from it.

If you're an ISP, it would be nice to give your customers that extra speed.

IPv6 in your data center also means your security team has an easier 
time tracking down miscreants than if they were behind CGN. Any security 
tool without IPv6 is blind to 54% of US traffic, 24% of CA traffic, 27% 
of global traffic.

Renumbering into IPv6 might mean you can make addresses available for 
sale, and prices are approaching the point where that makes sense.

For ISPs, you should absolutely figure out your IPv4 run rate, i.e., 
when you'll run out of IPv4 addresses. Then the PHBs have to decide what 
to do about that: deploy IPv6 and hope it's a viable alternative (with 
translation?), buy IPv4 addresses (at today's prices or tomorrow's, and 
how many addresses?), or deploy NAT44 and hope customers are okay with it.

For ISPs, consider how many of your customers are medium to large 
companies. These customers may need IPv6, either to sell their own 
addresses, or to connect with branches or partners who are out of IPv4. 
There are ISPs in the world who only support native IPv4 because some of 
their customers can't get approval for IPv4 from US corporate HQ. Of 
course, they pay more for that. For that matter, consider how much you 
charge for additional IPv4 addresses, and the rate at which customers 
could decline that service.

(But wait, you say, PHBs don't want to lose the IPv4 revenue! Depends on 
whether the competition is likely to offer the cheaper alternative)

Finally, the Rabobank argument: Maybe there aren't important sites, 
tools, or architectures that are only available over IPv6 right now. 
When will there be? Five years? Ten? (Seven? How long will it take you to be 
completely IPv4-independent, and will it be done in time?

So there's an 8-slide deck for you. Good luck with that pitch! I'm 
interested in what feedback/pushback you get.


> b.

More information about the NANOG mailing list