Annotated Bibliography on Global States and Times in Distributed Systems

A distributed system consists of a collection of concurrently executing processes that do not use shared memory; processes communicate with each other solely through message-passing and coordinate among themselves to achieve a common goal. In such systems, global state descriptions are not readily available, and a common time base does not exist. However, the notion of global states and snapshots lies at the core of many problems in distributed computation, and, as Lamport pointed out in a classical paper, the concept of time is fundamental to our way of thinking. Example applications of the global states and snapshots are distributed algorithms for detecting stable properties which hold during a computation, such as deadlock detection, termination, checkpointing and recovery in distributed databases, and monitoring and debugging of a distributed program. Because of their fundamental nature, a signicant body of research results about both global states and global times has been developed in the past few years. This bibliography identies these research results and provides brief annotations. It deals with the need to take snapshots of global states in a distributed system, and the need for global time obtained through one of the following two methods: synchronizing local clocks with reference to real time, and building logical clocks and virtual time without reference to physical or real time