Course description:
Nowadays, the IT infrastructure of many companies and institutions includes modern compute clusters. The procurement, operation and the efficient usage of such parallel systems introduces new and complex requirements.
To address these emerging requirements, the High Performance Computing Center of the University of Stuttgart (HLRS) offers a vendor-independent workshop. Topics span from the design of compute clusters to details on different hardware components, operating systems, file systems and modes of operation, touching on specific software solutions. Further, typical problems and strategies for their solution will be discussed.
HLRS has been running compute clusters for many years, providing compute resources for scientific and industrial simulations. HLRS is in constant dialogue with users and hardware providers and has accumulated a large knowledge base in cluster computing.
Programme
The agenda of the workshop is outlined below and is designed to leave room for dialogue between the participants and the attending domain experts. Note that this course will be taught in German.
First day:
Description of Hardware Concepts and Architectures
a. Cluster Structures
b. Network
c. Storage, file systems
d. Requirements for infrastructure
e. Performance
f. Tour through 3-D visualization CAVE and computer room
Second day:
System Software, Middleware, Tools
a. Operational concept
b. Operating systems
c. Storage concept, backup
d. Monitoring
e. Batch systems
f. Visualization
g. Parallel Programming
The course language is German.
Application procedure:
See http://www.hlrs.de/training/2018-06-20-cluster/ for full course details including registration procedure.
Note that attendees must pay a fee to attend this course.