5 Two basic models of distributed computation

Distributed computing and parallel computing study different kinds of systems, but they both study concurrent algorithms that run on collections of processors. In parallel computing, processors are tightly coupled, which usually means they move in lockstep, communication between processors is fast, and neither processors nor communication channels suffer from errors. By contrast, in a distributed system, processors are loosely coupled, which often means that different processors take steps at different times, messages placed in communication channels may take a while to reach their destinations, and processors and message channels may fail completely.