SEMANTICS OF RECOVERY LINES FOR BACKWARD RECOVERY IN DISTRIBUTED SYSTEMS

This paper addresses the definition of recovery lines in the context o f backward recovery whose aim is to cope with failures in distributed sytems. A general framework that allows for several semantics of recov ery lines is introduced. Key notions such as missing messages and orph an messages are precisely defined and their impact on the definition o f consistency of recovery lines is carefully analyzed. Basic mechanism s such as local checkpointing, messages identification and (optimistic or pessimistic) messages logging are then discussed as an illustratio n of (coordinated or uncoordinated) checkpointing protocols.