A generic log-service supporting fast recovery in distributed fault-tolerant systems

Logs are an important facility for fault-tolerant distributed systems since they allow to reliably store information that is needed to provide a global consistent system state also in the presence of failures. The authors focus on the problem of fast recovery after a node crash. The approach is mainly based on minimizing the number of log records to be retrieved. This is achieved by periodically discarding obsolete information in a very efficient manner without effecting the normal logging procedure. The main idea behind is that the generic Log-Service provides a high level interface to the application which allows the Log-Service itself to interpret the semantics of log records without consulting the application during run-time. In addition, the authors are able to reduce the overhead in analyzing the log contents during restart by scanning the log only once and only forward.