Computer Science Colloquium
High Performance Computing @ BMWMon 05.05.2008, 16:30, 60 minutes
AbstractA typical supercomputer is a dedicated machine bought for a specific task with a homogeneous system architecture. This concept is also reflected in the structure of the TOP500 list. Until 2006, BMW Group implemented HPC systems in this typical way. There were separate systems to satisfy business requirements. To be prepared for the future a new concept had to be set up. The aim was to define a cost efficient scalable infrastructure with the following design targets: robustness, high availability, performance, flexibility, maximum utilization and scalability. The result is a multi purpose cluster based on commodity hardware. The solution covers all aspects like servers, interconnect, security, cluster management and job flow. The cluster management layer is the key for optimal integration into the existing complex worldwide CA infrastructure and basis for optimized job flows. A sophisticated queuing system implementation is the factor of success for optimal workload to gain highest utilization and efficiency. The flexibility concept made it possible to double the capacity since 2006 in three stages of expansion without interruptions and technological restrictions. Further expansions and technology updates are planned.
BioThomas Hagauer attended a Higher Technical College in Steyr, Austria. During his studies he worked as an assistant at IAIK (Institute for Applied Information Processing and Communications, Graz University Of Technology). He graduated from the Technical University in Graz, Austria, in telematics in 1998. After one year at IBM he joined BMW where he came in touch with CAE Infrastructure. Until 2004 he was responsible for CAE Infrastructure supporting the Diesel Engineering Centre. From 2004 to 2006 he was responsible for Operations HPC and CAE Clients (BMW Group) and in 2006 he joined the department for IT Solutions and Technology Standards.
Invited by o. Univ.-Prof. Dr. Jens Volkert
The Computer Science Colloquium is organized by the Department of Coputer Science at JKU, the Österreichische Gesellschaft für Informatik (ÖGI) and the Österreichische Computergesellschaft (OCG).