System management framework and tools for Beowulf cluster

As the system size of Beowulf-Class cluster system scales up, system administration becomes challenging due to the increased complexity of the system. The lack of powerful tools for the administrative tasks even makes the matter worse. This paper presents a proposed framework for system management of large Beowulf cluster that divided the management task into well defined layers of service. A cluster management tool called SCMS (SMILE Cluster Management System) has been developed according to this framework. SCMS provides a rich set of services and tools for large cluster administration. The current implementation of SCMS is discussed referenced to the proposed framework. The software is now freely available on the Internet.