Pekka, I think there is an overarching issue about deployment/transition that relates to the NAT thing; providing a new service/feature/capability by only adding one box to the network is a lot easier sell than for instance - requring a box in the peer's network - requring all the routers in the path to do something new - requring the ISP to do something new
Thus even though NATs cause lots of problems, many would be swayed by the low entry cost of just adding a box to get some particular new capability. And NATs are used as a technology as part of providing different user visible capabilities such as - "connection sharing" from home - load balancers (without needing any support in the servers behind the LB) - multihoming boxes (some combining NAT and a DNS server for inbound and outbound multihoming support without ISP involvement or host awareness) It isn't clear to me at what point the pain caused by NAT in different cases will be high enough to motivate a transition to some different technology, or whether there must be new capabilities (such as the ability to have multiple servers at the same port number in a home) that are perceived important enough to cause migration away from NAT solutions. Erik -------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------