One way I've heard it: DisruptionTolerantNetwork is 'DelayTolerantNetwork' as understood by the military... who isn't all that tolerant of '''delay''', but does understand that things get blown up or jammed.

A DisruptionTolerantNetwork is one that can continue providing degraded service in the event of disruption (i.e. robust), and that can restore service when the disruption is alleviated (i.e. resilient).  Common problems in DisruptionTolerantNetwork are that nodes can be isolated, destroyed, and that parts of the network can become segmented, after which point communications may slow significantly (potentially operating upon some major information dumps when working with mobile nodes).  Degraded service may imply that, for example, one is forced to go from tele-conference to only providing recorded messages to be delivered when intermittent connectivity is established, or when a vector can be found to transmit the information.  As a consequence of this focus, much of the study of DisruptionTolerantNetwork''''''s is actually the study of various D''''''isruptionTolerantService''''''s and graceful degredation and restoration thereof.  However, just as much is focused upon identifying efficient mechanisms for delivering information even in a segmented network (e.g. using a flying plane as a mobile node only when you can tell it's headed in the right direction based upon the cooperation and triangulation from a wide spread of wireless systems).

While it does possess a military appeal, study DisruptionTolerantNetwork''''''s (and services) are quite valuable anywhere that network connectivity may be intermittent... including the wilderness (esp. the mountains and caverns).  Well... actually, that is of military appeal, too.  However, it doesn't quite bring to mind the exciting image of taking an RPG to a Cisco router.

The study of DisruptionTolerantNetwork''''''s and DelayTolerantNetwork''''''s isn't entirely new, but it's still more based in research than in application.  This is a somewhat unfortunate consequence of momentum in the existing structure... the same sort of reason that IPv6 has yet to be adopted.  New networks benefit more from this research, such as provision of Internet to vehicles on-the-move... and, of course, military applications of the research will help drive the technology.