Distributed Computing
Effects of message loss on the termination of distributed protocols
Information Processing Letters
Distributed snapshots: determining global states of distributed systems
ACM Transactions on Computer Systems (TOCS)
Time, clocks, and the ordering of events in a distributed system
Communications of the ACM
Knowledge and common knowledge in a distributed environment
PODC '84 Proceedings of the third annual ACM symposium on Principles of distributed computing
A knowledge-theoretic account of negotiated commitment
A knowledge-theoretic account of negotiated commitment
Hi-index | 0.00 |
We identify new circumstances under which processes in faulty distributed systems must communicate for one process to gain knowledge about another. Our main result says that, in systems with process crash failures, message loss, or asynchronous processes, if a proposition of a certain type about a process p does not hold, and later another process q knows that the proposition holds, then there was a message chain from p to q. Systems in which processes vote, bid, or transmit private values are often ones in which processes gain knowledge of propositions of the sort described in our result. One can use this result as a new tool in showing message lower bounds and impossibility results and in designing protocols. We demonstrate this by showing a new impossibility result for commitment problems: if a round-based commitment protocol is resilient to process failures and recovery and such that a message may be received only in the round in which it is sent, then the protocol may run forever.