Design, Implementation, and Verification of the Reliable Multicast Protocol

摘要

This document describes the Reliable Multicast Protocol (RMP) design, firstimplementation, and formal verification. RMP provides a totally ordered, reliable, atomic multicast service on top of an unreliable multicast datagram service. RMP is fully and symmetrically distributed so that no site bears an undue portion of the communications load. RMP provides a wide range of guarantees, from unreliable delivery to totally ordered delivery, to K-resilient, majority resilient, and totally resilient atomic delivery. These guarantees are selectable on a per message basis. RMP provides many communication options, including virtual synchrony, a publisher/subscriber model of message delivery, a client/server model of delivery, mutually exclusive handlers for messages, and mutually exclusive locks. It has been commonly believed that total ordering of messages can only be achieved at great performance expense. RMP discounts this. The first implementation of RMP has been shown to provide high throughput performance on Local Area Networks (LAN). For two or more destinations a single LAN, RMP provides higher throughput than any other protocol that does not use multicast or broadcast technology. The design, implementation, and verification activities of RMP have occurred concurrently. This has allowed the verification to maintain a high fidelity between design model, implementation model, and the verification model. The restrictions of implementation have influenced the design earlier than in normal sequential approaches. The protocol as a whole has matured smoother by the inclusion of several different perspectives into the product development.