Abstract: We witness an increasing need for integrated software ecosystems that will need to combine current “island” ecosystems for traditional supercomputers, cloud-based and edge-based systems and bridge the gaps between them. These integrated ecosystems must facilitate the full life cycle of edge-cloud-HPC continuum workflows, including initial modelling, programming, deployment, execution, optimization, as well as monitoring and control. It will be important to ensure adequate reproducibility of workflow results and to find ways for creating and managing trust when sharing systems, software and data.
This joint talk will discuss challenges posed by the emergence of the computing continuum and introduce some solutions proposed by the KerData team at Inria. In particular, we will present our ongoing work on enabling reproducible experiment execution across the continuum, how to provide adequate support for ML-oriented workloads, as well as innovative approaches to dynamic storage allocation.