Selon "Manfredi, Albert E" <[EMAIL PROTECTED]>: > I'm not sure what different the intermediate routers would have to do, > if anything.
Nothing. But policy boxes and firewalls would be slower because of the IHL difference. > It's a way of tackling the > address space problem without depending on a new version of IP or even > on a new address format. It DOES depend on a new addres format. You'd not only have to define a way to write host A behind NAT B, but you have to use a new sockaddr addresses family so that this actually become usable in any application willing to leverage this function. And you need a new DNS record type. Besides it's obviously broken if you have more than one level of NAT because sockaddr and DNS records would have a fixed length, so you couldn't put this IP option multiple times. And, it would cause a bunch of fragmentation headaches on the NAT box to insert the IP option. -- Remi Denis-Courmont http://www.simphalempin.com/home/ -------------------------------------------------------------------- IETF IPv6 working group mailing list ipv6@ietf.org Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------