Toward globally optimal resource management in large-scale real-time distributed computer systems

This paper discusses the issues and promising approaches in obtaining rigorous specifications of the quality-of-service (QoS) requirements associated with application functions and using such specifications as the main driver for cost-effective resource allocation. The application environments in which component failures must be handled with sufficient efficiency in order not to miss the application objectives and also execution resource requirements of various application functions can change dynamically, are considered. The importance of structuring various resource users, i.e., application software components and middleware components, in easily analyzable forms, is emphasized. A promising middleware architecture that supports application software structured in real-time object-oriented forms and also manages execution resources for fault-tolerant system operations, is presented