Re: [v6ops] I-D Action: draft-ietf-v6ops-6to4-to-historic-06.txt - alternatives to 6to4

[Jeroen Massar <[e-mail]>]

On 2014-11-04 01:12, Keith Moore wrote:
> On 11/03/2014 06:53 PM, Owen DeLong wrote:
>>
>>> On Nov 2, 2014, at 9:09 PM, Keith Moore <[e-mail]
>>> <mailto:[e-mail]>> wrote:
>>>
>>> On 11/02/2014 11:35 AM, Metzler, Dan J wrote:
>>>> That said, back to the original discussion, involving protocol 41,
>>>> if AT&T is truly blocking protocol 41 because they provide IPv6
>>>> native support across the board already, that seems a lot less
>>>> unreasonable.
>>> protocol 41 isn't just IPv6 over IPv4.
>>>
>>> Keith
>>>
>>
>> What else is it? GRE is Protocol 47.
> 
> I thought protocol 41 was generally used for IP (of whatever version)
> encapsulated in IP (of whatever version).

No. If you see a value of '41' in a "protocol" or "next header" field
then it means the next header/protocol is IPv6.

For IPv4 in IPv6(or something else) you would use a next header of 4.

eg 4in4 would be [IPv4, protocol = 4][IPv4, protocol = TCP/UDP/etc]
eg 6in4 would be [IPv4, protocol = 41][IPv6, next header = TCP/...]
eg 6in6 would be [IPv6, next header = 41][IPv6, next header = TCP/...]
eg 4in6 would be [IPv6, next header = 4]][IPv4, protocol = TCP/...]
etc

> IANA's protocol number
> registry seems to cite RFC 2473, which is about encapsulating various
> kinds of things in IPv6  (though admittedly the distinction between the
> different registries isn't very clear).

Nack. 2743 is effectively about 6in6.

If you peek at:
https://www.iana.org/assignments/protocol-numbers/protocol-numbers.xhtml

You'll see it points at the similar number 2473.

Note that the middle two digits are swapped with the previous one, which
is likely where you got that honest mistake from.

And as most people call 6in4/tsp/ayiya/etc "IPv6 tunnels" 2743 does not
make this terminology much easier either ;)

> Of course 6to4, 6over4, 6in4
> all use protocol 41 and all encapsulate things over IPv4.

Don't forget about 6rd and all other protocols, eg AYIYA who indicate a
'next header' field of 41 to indicate IPv6 packets following there.

> What I
> haven't found is a use of protocol number 41 to encapsulate IPv4 in
> something else, so I might have been mistaken about that.

Because 41 is not meant for that. You need '4' for that (RFC2003).

Fun point there is that there is also 94 aka "ipip" but that is a
completely different beast again that is afaik nwhere to be found how it
is actually used.

Oh, yes all of this is fun and confusing ;)

(Not even going to GRE which uses the 802.11 'next header' numbers...)

Greets,
 Jeroen