[Bp_ipv6] Can we do the math? ( was [MEAC ICANN] Fwd: Interesting IPv6 metrics)

Marco Hogewoning marcoh at ripe.net
Fri Aug 26 11:06:44 EDT 2016


>> On the predicted churn when deploying NAT444, where would a customer go? To the other carrier that uses NAT444?
> 
> Ha!  Assumes all carriers transition at the same time--I wish that were true!

Okay, you might find the old incumbent can still provide you with regular (native) IPv4. I was particularly referring to those markets where 0 IPv6 is observed.

>> 
>> I dunno, happy to hear other people’s thoughts on how big of a thread NAT really is to day-to-day use of common applications and equipment?
> 
> My choices as an ISP are:
> * Buy addresses. Gets very expensive very quickly.
> * Deploy CGN. Extends the runway, but I gradually have to move all of my customers through CGNs. Maybe not as expensive as it once was, but I'm still buying boxes to rewrite headers on every single packet, which is expensive in processing power.
> * Deploy IPv6 plus something to reach legacy IPv4. Provides IPv6 as a relief valve for anything that doesn't like CGN, plus my need to rewrite packet headers decreases over time, not increases. MAP and DS-Lite provide connectivity between IPv4 CE and IPv4 content, but only if the gateway supports it; that makes a nice transition plan, and means an IPv6-only access network in one CE life cycle. NAT444 provides connectivity between IPv4 CE and IPv4 content, and requires no support on the local gateway. 
> 
> The final option is to say, "We're not growing, we don't need to do anything!" I call those cases business opportunities for competitors.

Useful input

> All you had to do to flush me out of lurk mode was to troll me with "CGN isn't so bad!" I'm so predictable.

Thanks for that,

Marco



More information about the Bp_ipv6 mailing list