Model-Driven Resilience Assessment of Modifications to HPC Infrastructures

Abstract

The increasing influence of surrounding factors on HPC infrastructure operations, e.g., increasing cost pressure or evolving power consumption, requires an analysis of the indented effects of modifications in the context of the entire HPC infrastructure and its capabilities, before this modification is further (costly) explored or executed actually. Therefore, a meta model is provided that enables a domain-spanning capability effect analysis, which respects the complex component interplay of contemporary HPC infrastructures. The meta model in turn is used as basis for the model-driven development of tooling support.