A Method for Evaluating Efficiency of Protocols on the Asynchronous Shared-State Model

Abstract

Distributed systems are commonly modeled by asynchronous models where no assumption is made about process execution speed. The asynchronous model is preferable to the synchronous one because the model reflects the fact that a distributed system consists of computers with different processing speeds. However, the asynchrony of the system makes it difficult to evaluate efficiency (performance) of distributed protocols. This paper defines a class of distributed protocols called linear state-transition protocols, in the state-communication model, and shows that efficiency of such protocols in the asynchronous distributed model can be derived from analysis of their synchronous execution, where all processes are synchronized in the lock-step fashion. This provides an effective method for evaluating efficiency of the linear state-transition protocols in the asynchronous distributed model. The paper also demonstrates the effectiveness of the method by applying it to the self-stabilizing alternator.