The aim of this deliverable is to provide indications about how to use container to improve portability of applications on HPC systems.
Important steps to run a containerized application in a HPC system include the combination of a container instance within the job batch scheduler available on the system. In the first sections, this deliverable is focused about the usage of MPI libraries, network interconnection drivers and accelerators within a container. Then, some considerations about the integration with workload managers are provided, together with an optimization for Singularity containers.
Finally, a case study about containerization in cloud environment is reported.