el/REL: a family of reliable multicast protocols for distributed systems

Abstract

A reliable multicast service that ensures atomic delivery, such that a multicast is completed successfully despite intervening failures (for example, the crash of the sender during a multicast) is highly desirable for building dependable distributed systems. This paper presents a family of multicast protocols that are easy to understand and implement and are capable of providing such a service with differing message ordering requirements. These protocols provide good performance despite intervening node crashes, thus making them suitable for systems requiring timely responses in the presence of component failures.