"(Marc Riese)" <riese@litsun.epfl.ch> (04/09/91)
I am a student working on network diagnosis and I am looking for descriptions of difficult network problems whose diagnosis involved some degree of reasoning about time and/or space. That is, in order to solve the problem, the diagnoser had to reason explicitly about when and/or where certain events happened that led to the problem. For example, a common problem on data-comm networks is `broadcast storms', where unanticipated or unintended transactions cause multiple broadcasts which get out of control and swamp the system. Udi Manber described other examples in "Chain Reactions in Networks" (IEEE Computer 10/90). I am looking for other examples, datacom or telecom, generic or specific. Any information or pointers much appreciated. Discussion welcome. Marc Riese Swiss Federal Institute of Technology