dslreports logo
 
    All Forums Hot Topics Gallery
spc
uniqs
15
InvalidError
join:2008-02-03

1 recommendation

InvalidError to rolande

Member

to rolande

Re: IPv6 non connect

said by rolande:

We won't see major transition, until there is that killer app that is only accessible via IPv6.

With IPv4 pools depleted, it is only a matter of time before ISPs start having to CG-NAT their subscribers.

I suspect people's interest in IPv6 is going to get much stronger once ISPs start charging extra for public IPv4 addresses or only include a public IP with their higher-end plans.

rolande
Certifiable
MVM,
join:2002-05-24
Dallas, TX
ARRIS BGW210-700
Cisco Meraki MR42

1 recommendation

rolande

MVM,

said by InvalidError:

With IPv4 pools depleted, it is only a matter of time before ISPs start having to CG-NAT their subscribers.

Everyone has been claiming the Internet sky is falling and that IPv6 transition must happen now, since 1995. While I don't disagree that the transition needs to happen, it still won't happen as fast as everyone thinks it will. Back in 1995 when the sky was first falling, NAT was introduced and classless routing, and suddenly the address scarcity problem with IPv4 was "resolved" or at least delayed considerably.

The next iteration of that kind of band-aid solution has been introduced by Cisco which is a stateless mapping of IPv6 to duplicate IPv4 addresses called MAP-T. So, there is an optional path available for those carriers stuck in the address starvation problem to continue supporting native IPv4 on their CPE, even after they have run out of unique IPs to hand out. Everyone knows that CG-NAT only scales so far before the hardware is crushed via memory starvation. If stateful translation were required, it would force native adoption much faster within the ISPs who were address starved. MAP-T gives them another out to ride out the transition period.

So, once again, the sky really is not falling. It will take time to truly reach critical mass where there is enough native IPv6 support, enough demand from IPv6 only clients, and content providers who see economic value in spending the money to fully transition their services. The adoption curve will look more like a hockey stick. It will be slow and steady adoption, until some point of critical mass occurs between native client population, native service provider support, and native content providers. At that point the adoption rate will skyrocket to nearly 100% in a fairly short amount of time relative to the previous rate over the past 10-15 years. I still think the knee of that curve is closer to 5 years out.