Stephen,
I've no idea what you think you mean when you say 'moving beyond
mandatory to implement'. My take is that encryption should never be
mandatory to implement.
MTI security is what's called for by BCP 61.
...only for standards-track protocols.
And that document asks 'is encryption a MUST' and sensibly answers 'no'
as far as confidentiality goes. draft-farrell-perpass-attack goes beyond
that to insist on confidentiality everywhere.
So, your take on MTI seems far far broader than that in BCP 61.
[..]
I did not discuss processing cost, though that's also relevant,
particularly for low-end systems.
My concern is that the cost to designers of trying to design
transport protocols becomes unacceptably high, as any attempt
to design a transport protocol becomes defending against the design
of what has just become a security protocol.
I also meant that. Arguing that we don't know how to engineer
this is also a fine 1990's era argument.
DTNRG (late 2000s, yet still persisting as a group as e.g. the TCP
convergence layer isn't published yet) is a much more recent example
of this.
I view the failure of DTNRG work - where tailoring for the DTN
scenario was dropped in favour of emphasising security work, and
problems with the protocol would be fixed by the security framework,
but weren't - as an example of this. And where the IRTF leads,
the IETF follows.
Your and my descriptions of about 90% of things related to DTN
seem to be at odds. And that's the case here too.
Damn right.
Lloyd Wood
http://sat-net.com/L.Wood/dtn