ComputeOps: containers for High Performance Computing

• Main Authors: Cavet Cécile, Bailly-Reyre Aurélien, Chamont David, Dadoun Olivier, Dehne Garcia Alexandre, Guérin Pierre-Emmanuel, Hennion Pascale, Lodygensky Oleg, Marchal-Duval Gérard, Medernach Emmanuel, Mendoza Victor, Pansanel Jérôme, Randriatoamanana Richard, Sartirana Andrea, Souchal Martin, Tugler Julien
• Format: Article
• Language: English
• Published: EDP Sciences 2019-01-01
• Series: EPJ Web of Conferences
• Online Access: https://www.epj-conferences.org/articles/epjconf/pdf/2019/19/epjconf_chep2018_07004.pdf

The High Performance Computing (HPC) domain aims to optimize code to use the latest multicore and parallel technologies including specific processor instructions. In this computing framework, portability and reproducibility are key concepts. A way to handle these requirements is to use Linux containers. These "light virtual machines" allow users to encapsulate applications within its environment in processes. Containers have been recently highlighted because they provide multi-infrastructure environnement for both developers and system administrators. Furthermore, they offer reproducibility due to image building files. Two container solutions are emerging: Docker for micro-services and Singularity for computing applications. We present here the Com-puteOps project which investigates the container benefits for HPC applications.