extending at the edge

Simon Lockhart simon at slimey.org
Thu Oct 11 11:39:02 CEST 2012


On Thu Oct 11, 2012 at 05:55:28PM +0900, Lorenzo Colitti wrote:
> No, I think IPv4-style NAT is precisely what he means. Think about it this
> way: if a provider wants to charge on a per-/128 basis, and the user only
> wants to pay the provider once but connect multiple devices, then the only
> way to do that is IPv4-style NAT. As I noted earlier, in this model
> (almost) everybody is worse off, but it's not clear that there's a way to
> avoid that happening.


I know this if veering rapidly off ops, and onto business/commercials, but
an interesting topic. Hotspot, Hotels, Wireless providers want to be able to
charge on a "per subscriber" basis (and I'm not sure that you can move away
from that in these environments). This is typically done as a "one mac, one
subscriber" solution. In an IPv4 world, NAT is a way around this, and it's
tolerated, but not encouraged by the service providers. If they try to stop
it, it's done by traffic pattern or signature analysis. The service provider
doesn't need to do anything to enable end users to use NAT.

Now, in an IPv6 world, we don't really have NAT (yet), so people are trying
to work out how to get around this. By using DHCP-PD, it's making it obvious
to the service provider that you're sharing your connection with other devices.
Do they want you to do this? How do they know all these devices you want to
share to are your own, and you're not removing revenue from them.

I think at the end of the day, this is a business decision, not a technical
decision.

Simon


More information about the ipv6-ops mailing list