Re: TASL
Dave Crocker <[email protected]>
| Newsgroups | gmane.ietf.multi6 |
|---|---|
| Organization | Brandenburg InternetWorking |
| Message-ID | <[email protected]> |
Brian, BEC> Seriously, there is obviously a very close relationship (the actual BEC> protocol exchanges would be like those in MAST and NOID, so I didn't BEC> spell them out). But when I re-read MAST I don't see it clearly stated that BEC> a separate, reliable, and secure transport connection is used to carry the separate: yes. reliable: the spec calls for retransmission with exponential backoff. secure: the current spec calls for a random string, to protect against obvious within-session spoofing. doing more than this requires carefully deciding among tradeoffs for cost and benefit. TLS is an easy, reasonable choice, but the overhead of a tcp connection gets a little scary when the scaling effects of having every host-pair do this is considered. BEC> exchanges - that is suggested as an option at the end of the MAST security BEC> considerations, but it is basic to TASL. And I don't see any need to use BEC> XMPP as you suggest in MAST. So I'd argue it's a simpler approach, but BEC> clearly closely related. The question is whether there is need for a dynamic rendezvous service. It is offered as an adjunct, not core, capability in MAST, in order to support situations with mobile servers. d/ -- Dave Crocker <mailto:[email protected]> Brandenburg InternetWorking <http://www.brandenburg.com> Sunnyvale, CA USA <tel:+1.408.246.8253>, <fax:+1.866.358.5301>