> But in the IP world, there is a full continuum of states in between. Some > of these are candidates for a useful service, and some of which aren't. > - Allowing a higher packet drop rate across all the "lower priority" calls. Note that this scenario gives so-called *IMPORTANT* traffic the priority that is *DESIRED* while still allowing other traffic to get through. In a disaster scenario like the New Orleans hurricane, the official command and control structure was not the sole organization doing useful work. In such a chaotic scenario, common in war and in disasters, it would be unwise to only have the option of shutting off everybody else's service by preemption. The low priority traffic could still be carrying mission critical messages that are crucial to dealing with the disaster scenario. This reminds me of an anecdote that a friend of mine recounted some years ago. During the a war in some African country, maybe Angola, the international phone lines were usually so noisy that it was impossible for the embassy to hold a conversation with their leaders back in Africa. Faxes were also virtually impossible to transmit. He solved this problem by installing a UUCP email system that communicated back to Africa using Telebit PEP modems. Unlike other modems of the day, the PEP protocol divided the potential bandwidth of the communications channel into 256 frequency ranges, tested each range and used as many ranges as were functional. This testing and choosing process was repeated periodically so that as the noise pattern shifted, PEP would search out the cleanest parts of the channel and use those. The bursts of data that succeeded in reaching their destination are analogous to packets and the channel seeking behaviour is analogous to routing/failover. The end result was a reliable email communications channel. It could and often did take many minutes to transmit just a few lines of text, but it did function and this allowed the embassy to function where before they were paralyzed. IP networks allow this same possibility, that of finding a low bandwidth end-to-end path amidst a cacophony of cross-traffic. The IETF should do some more education on how the existing protocol set could be used in disaster/war scenarios without needing any new features tacked on to it. --Michael Dillon _______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf