Towards formalising Erlang failure and failure detection

Abstract

Lately, more emphasis is being put on building fault-tolerant parallel systems. This fact can be clearly seen from the number of companies that are opting to develop their systems in Erlang; a parallel language which is renowned for its error handling capabilities. A sound understanding of a system's behaviour when errors occur is the key to developing truly fault-tolerant software. This dissertation investigates Erlang's error handling mechanisms so as to better understand how Erlang behaves in the presence of errors. A formal model is defined in order to provide a precise and unambiguous description of the behaviour of these mechanisms. The correctness of the model is evaluated by considering a number of Erlang programs and comparing the behaviour as described by the model with that of actual Erlang. Ultimately, the defined model is animated through an evaluator.