[address-policy-wg] AW: [ipv6-wg] DRAFT: RIPE Community Resolution on IPv4 Depletion and Deployment of IPv6

> This all comes down to economics.  Adding IPv6 capabilities 
> to CPE access devices costs money, and CPE devices are often 
> chosen purely on the basis of cost alone.  Ergo, IPv6 
> capability is bad for business, if you manufacture CPE boxen.

IPv6 is a software upgrade. Most of these boxes can be flashed in the
field and many of them are based on Linux or BSD, so in most cases it is
only a matter of including and testing the IPv6 code that already
exists.

> The real problem here is the lifetime of CPE devices.  I'm 
> going to estimate a rough half-life of 3 years.  This is 
> going to mean an awful lot of CPE access device swapouts to 
> move to teh ipv6 intarweb.

This means that if we can get manufacturers to include IPv6 during the
next 12 months, then in 4 years, half the CPE will support IPv6 through
pure attrition. That's not too bad. Once the kit is on the market, this
schedule can be speeded up if required.

Also, remember that IPv4 doesn't suddenly stop working. You could
reasonably expect to leave most existing customers alone and only deal
with CPE upgrades for the few that want to upgrade.

The CPE manufacturers can move pretty quickly if they see a demand for
IPv6.

--Michael Dillon