Recovery with limited replay: fault-tolerant processes in Linda

Research in the area of fault-tolerant distributed systems has focused to a large extent on data surviving various forms of failure. The replica control algorithms for maintaining mutually consistent replicas abound in number. However, comparatively little work has been devoted to making processes recoverable. In domains other than databases and transaction processing, faulttolerance generally implies both fault-tolerant data and fault-tolerant processes. In environments where cooperation among processes is important we argue that high availability of processes in addition to their recoverability is crucial. Our specific interest is in the Linda tuple space paradigm. In this paper we discuss efficient techniques for making Linda processes recoverable and outline some characteristics of Linda that make it particularly suitable for implementing fault-tolerance. We also propose a simple extension to our recoverable process mechanism that makes processes highly available. [keywords: fault-tolerant processes, high availability, recovery, Linda tuple space, replay, message logging].