Orchestration of Distributed LOFAR Workflows
The LOFAR radio telescope produces petabytes of data every year. Radio Astronomers use complex multi-step pipelines to process this data and produce scientific images.
- Mechev, A.P.
- 09 december 2019
- Thesis in Leiden Repository
The LOFAR radio telescope produces petabytes of data every year. Radio Astronomers use complex multi-step pipelines to process this data and produce scientific images. In this thesis, we describe methods for astronomers to efficiently process massive amounts of data on high-throughput clusters. We gain insights into efficient acceleration of complex scientific pipelines, as well as understanding of the scalability of the underlying software. These insights ensure the scientific output of LOFAR keeps up with the pace of the data the telescope produces. Moreover, our results can be applied to arbitrary scientific studies, greatly increasing the potential of the LOFAR telescope. Finally, our discoveries can be applied to current and future telescopes. These lessons will help astronomers efficiently producing science in the big data era.