The SysMES Framework: System Management for Networked Embedded Systems and Clusters

Automated system management for large distributed and heterogeneous environments is a common challenge in modern computer sciences. Desired properties of such a management system are, among others, a minimal dependency on human operators for problem recognition and solution, adaptability to increasing loads, fault tolerance and the flexibility to integrate new management resources at runtime. Existing tools address parts of these requirements however there is no single integrated framework which possesses all mentioned characteristics. SysMES was developed as an integrated framework for automated monitoring and management of networked devices. In order to achieve the requirements of scalability and fault tolerance, a fully distributed and decentralized architecture has been chosen. The framework comprises a monitoring module, a rule engine and an executive module for the execution of actions. A formal language has been defined which allows administrators to define complex spatial and temporal rule conditions for failure states and according reactions. These rules are used in order to reduce the number and duration of manual interventions in the managed environment by automated problem solution. SysMES is based on standards ensuring interoperability and manufacturer independence. The object-oriented modeling of management resources allows several abstraction levels for handling the complexity of managing large and heterogeneous environments. Management resources can be extended and (re)configured without downtime for increased flexibility. Multiple tests and a reference installation demonstrate the suitability of SysMES for automated management of large heterogeneous environments.