HIGHLY AVAILABLE COMMUNICATION.

Abstract

The authors describe some aspects of the design and implementation of a fault-tolerant distributed facility for communication (DCF) and location transparency. They emphasize novel protocols including clock synchronization and atomic broadcast, as well as the general adaptiveness and ease of use of DCF. The simple but powerful collection of fault-tolerant protocols provides a good foundation for extension to many distributed applications where fault tolerance, high availability, ease of use, and ease of growth are important