Hi Peter,
On Dec 5, 2008, at 9:20 PM, Peter Dambier wrote:
Without NAT66 they cannot route because IPv6 addresse dont aggregate and all those many tunnels keep address tables exploding.
This doesn't make any sense to me. Why don't IPv6 addresses aggregate? AFAIK, they aggregate the same way that IPv4 addresses do.
There are known issues with certain technologies (like using inter-ISP tunnels for site multihoming) and route aggregation, but it is my understanding that those problems exist in both IPv4 and IPv6. Is you ISP describing an IPv6-specific problem?
On the other hand if we can build a little something that can replace your existing NAT44 router and allows you to 6to4 through your dynamic IPv4 address but keeping your ULA or PI intact, that little something might be called NAT66 and two of them might be called symmetric NAT66.
It is my understanding that the term "symmetric NAT" is already in use to describe something different from what you are describing, so you might consider a different term to avoid ambiguity.
It is _exactly_ the point of NAT66 that a customer can use his ULA or PI address internally and his ISP-provided, aggregated address externally. I am not sure, though how the rest of your solution relates. Without a global mapping service of some type, it would not be useful for my ULA or non-globally-routed PI addresses to get to your site, because you couldn't send a packet back to those addresses.
If there is a draft to follow, the linux IPv6 guys and and the BSD guys will implement it. It will finally find its way into those little boxes that do connect you to the IPv4 internet today.
Sadly (or perhaps fortunately :-)), it doesn't always follow that the open source guys implement whatever we write. Hopefully we can have some influence in this case, however.
Margaret _______________________________________________ nat66 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nat66
