Software fault-tolerant distributed applications in LiPS

Abstract

This paper illustrates how software fault-tolerant distributed applications are implemented within LIPS version 2.4, a system for distributed computing using idle-cycles in networks of workstations. The LIPS system [SR92, SR93,STea94,Set95,SF96,ST96,SL97,Set97] employs the tuple space programming paradigm, as originally used in the LINDA programming language. Applications implemented using this paradigm easily adapt to changes in availability as they occur in workstation networks. In LIPS, applications are enabled to terminate successfully in spite of failing nodes by periodically writing checkpoints, freezing the state of a computational process, and keeping track of messages exchanged between checkpoints in a message log. The message log is kept within the tuple space machine implementing the tuple space and replayed if an application process recovers. This assumes deterministic behavior of the application process but allows independent checkpoint generation and alleviates the need for application-wide synchronization in order to generate sets of consistent checkpoints.