HPC Cluster Management with Open Source Software

• Autores: Tinetti, Fernando Gustavo; Rios, Leopoldo J.
• Año de publicación: 2017
• Idioma: inglés
• Tipo de recurso: documento de conferencia
• Materia: Ingenierías y Tecnologías; High Performance Computing; RBAC; Open Source Software; RRDtool; Ganglia
• Institución: Comisión de Investigaciones Científicas de la Provincia de Buenos Aires

Scientific institutions and laboratories have incorporated and maintain a significant amount of computing resources over the years. Research centers may have unused hardware facilities, with significant processing throughput that could be made available to other institutions. In this paper, we propose a software infrastructure built on existing and new open source software for sharing its computing infrastructure, in a programmed, monitored, and secure way. The Role Based Access Control (RBAC) model is considered a mature and flexible technology, and is a very popular paradigm today. It offers a more secure alternative to the all- or-nothing superuser model. With RBAC, it is possible to apply a specific security policy, and it means that a user has the necessary amount of privileges to perform a task. Taking advantage of the known features of RBAC, it is proposed to add specific HPC monitoring functionality. The HPC monitoring facilities are intended to aid users of scientific programs in their optimization and parallelization tasks of processing intensive jobs.