Can IPv4 tunneled in IPv6 (or vice versa) handle any traffic/protocol types?

[OpenSource Ghost]

If ISP enacpsulates IPv4 in IPv6 somewhere down the line (or vice versa), is it able to pass all traffic or are there protocols and traffic types that would hard to encapsulate or even break such encapsulation? For example, IPv6 has a hard time encapsulating packets with MTU less than 1280, but I think that applies only to IPv6-to-IPv6 connections.

 

[SamirD]

Interesting question. I don't think if something is ecapsulated properly (I'm thinking of ipsec vpn tunnels), that it would really matter as packets would be sized as needed.

 

[Nicklebon]

There is no "hard time." The minimum mtu in ipv6 is 1280 and in general is often the maximum. Assuming you're the one doing the encap, hard code your end to 1280 and you'll likely see no issues.

 

[toast0]

In theory, you can encapsulate whatever. Your MTU on the tunnel is outside MTU - tunnel overhead, and if your v6 MTU gets below 1280, you might have issues with some stuff. TCP should be fine, as you can set MSS less than MTU during the handshake. UDP won't be fine, unless the application protocol knows how to do it. Non-tcp/udp/icmp, good luck.

Also, if your ISP is doing NAT64 or whatever, it probably means you're sharing a IPv4 address with a bunch of other people, and good luck getting any inbound connections and non-tcp/udp/icmp is even less likely.