Re: I-D Action:draft-vandevelde-v6ops-harmful-tunnels-01.txt

[Brian E Carpenter <brian.e.carpenter@gmail.com>]

I am troubled by this draft. It makes broad assertions without giving
detailed scenarios, and it doesn't offer any solutions.

For example, the reference given for 6to4 is RFC 3056. As far as I
know, 6to4 as described in RFC 3056 has been deployed very little.
But that is a *managed* scenario, no so different from 6rd, which
requires 6to4 routers and relay routers to be configured according to
certain guidelines, designed to prevent exactly the sort of problems
this draft is complaining about.

What has been deployed reasonably widely is a completely different
scenario, host-based 6to4 relying on an anycast mechanism for finding
a relay router, as defined in RFC 3068, which isn't even referenced.
There are well-known problems with that model, resulting in one-way
connectivity in many cases. If that is what the draft wants to condemn,
that model should be described in detail and its failure modes given
in detail.

The same applies to Teredo. We need a more carefully described scenario
and an explanation of the failure modes.

Some positive proposals for how to correct these failure modes would
be more productive than a "considered harmful" conclusion, too.
6to4 exists to bypass IPv4-only ISPs; Teredo exists to bypass
IPv4-only CPEs; these are real world problems that need solutions.

Regards
   Brian Carpenter