Deployment of a programming framework based on microservices and containers with application to the astrophysical domain

Abstract

Scientific research is becoming increasingly global and has to address challenges that cannot be handled by small groups or individual researchers. Global research initiatives are often facing the lack of cooperative computing frameworks and storage resources. Such frameworks and resources are needed to address scientific problems at large scales. This paper investigates the potential of using container technologies over distributed computing infrastructure in the astrophysics domain. Due to the complexity of astrophysics applications, it is necessary to enable a parallelism pipeline for their execution. Such applications require High-Performance-Computing (HPC) environments to enable easy and reproducible deployment of astrophysics applications across multiple HPC cites “BRANE”. BRANE provides easy and transparent access to end-users, allowing them to define their execution pipeline.11https://wiki.enablingpersonalizedinterventions.nl/user-guide/. To validate the proposed solution, we considered an astrophysics application that uses the XSPEC package for spectral analysis from HEASoft. XSPEC software is used for spectral fitting of data from X-rays telescope observations. The Xrays data are very large that exceed 1Gb, and obtaining such a huge amount of data requires awesome time and human efforts. The conducted experiments have shown the usefulness, portability, and reproducibility of the proposed solution.