Practical application and implementation of distributedsystem-level diagnosis theory

摘要

A DSD (distributed self-diagnosing) project that consists of theimplementation of a distributed self-diagnosis algorithm and itsapplication to distributed computer networks is presented. TheEVENT-SELF algorithm presented combines the rigor associated withtheoretical results with the resource limitations associated with actualsystems. Resource limitations identified in real systems includeavailable message capacity for the communication network and limitedprocessor execution speed. The EVENT-SELF algorithm differs frompreviously published algorithms by adopting an event-driven approach toself-diagnosability. Algorithm messages are reduced to those messagesrequired to indicate changes in system those messages required toindicate changes in system state. Practical issues regarding the CMU-ECEDSD implementation are considered. These issues include thereconfiguration of the testing subnetwork for environments in whichprocessors can be added and removed. One of the goals of this work is toutilize the developed CMU-ECE DSD system as an experimental test-bedenvironment for distributed applications