Updates on Yandasoft/ASKAPsoft Deployment using Docker and Singularity
Dr Paulus Lahur1
1CSIRO, Clayton, Australia
In this presentation we would like to share our experience in deploying a complex scientific software to users, who might have a variety of computing hardware, ranging from laptop to supercomputer. Specifically, we would like to discuss the use of container technology.
Our team in CSIRO’s Space and Astronomy business unit is developing a suite of radio astronomy software called Yandasoft, which is intended for the calibration and imaging of interferometric radio telescope data. It was originally developed as ASKAPsoft for the Australian Square Kilometre Array Pathfinder (ASKAP) project. The software has been released as open source, and it has gained interest from users in a number of research groups in the field of radio astronomy. Container technology is essential for the deployment to end-users, because the software and its dependencies is downloaded as an integral tool, ready to run on hardware commonly used for scientific computing.
In our deployment workflow, we generate a set of Docker images that contain the software as well as a variety of parallel computing implementations. Users can then download a suitable image either using Docker or Singularity, and immediately ready to run it without having to install the software. Both users and developers have gained significant benefit from this approach, as the installation step can usually be quite frustrating. The latest updates on the techniques that we use to achieve this will be presented.
Currently the presenter works as research software engineer in Scientific Computing in Information Management and Technologyy … (IMT) in CSIRO, where he helps various research projects in DevOps for scientific software. Previously he worked in Japan Aerospace Exploration Agency (JAXA) as researcher in Computational Fluid Dynamics (CFD)