Distributed Cosmic Ray Detection Using Cloud Computing

This article presents a distributed computing approach to detect cosmic rays in images taken by the Hubble Space Telescope (HST). A cloud computing implementation is developed to improve the overall processing time for the available images dataset (15 TB), containing dark images from several HST instruments. A specific architecture is presented where images are stored in a replicated and highly available storage system. Image processing is performed on virtual machines from the Azure Batch framework using a developed Python application. The experimental evaluation shows that the architecture accomplished the purpose of processing the complete dataset based on scaling computing resources in terms of processing nodes. Speedup improved in a factor of \(6.57{\times }\) over a previous implementation using Apache Mesos. The overall computation took 10 days to complete and results are stored on a non-relational database available to astronomers and researchers.