CompDaF

The need for remote astronomy visualisation is becoming more demanding as time passes. As infrastructure improves and continues to exceed expectations, there is a need for a system that can accurately analyse and render the images produced from the array telescopes. The CARTA system is a tool used to process this image data, and new methods are being researched to deal with this exponential rise in data size. The dataflow model is a possible solution to this issue, with a scalable architecture that can run efficiently across massively distributed clusters, while eliminating the issues inherent in parallel programming.

This project presents two prototypes of the dataflow model using two different frameworks, namely Raftlib and DALiuGE. Both prototypes have been subjected to thorough efficacy and efficiency testing with results being compared against CARTA’s current backend. Both implementations showed satisfactory performance with the results being valid and explainable. We conclude that the dataflow library is indeed a viable alternative to replace certain features of the CARTA backend.