Performance evaluation of an integrated RFI database for the MeerKAT/SKA radio telescope.

For radio telescopes, radio frequency interference from terrestrial and other sources is a recognized problem that contaminates the signal (RFI) and must be tracked and ultimately removed. At the MeerKAT/SKA telescope, RFI is recorded with a variety of devices, including telescopes, sensors, and scanners; but the combination of data from these multiple sources to yield a unified view of RFI remains a challenging problem. Previously, we demonstrated that a scalable database model with an implementation based on the Polystore framework is a potential solution for RFI monitoring. Here we extend this work, implementing the database model in an integrated environment and evaluating its performance across a range of workloads with three data stores: SciDB, PSQL, and Accumulo. We find that SciDB and Accumulo scale better than PSQL under multi-user environments. Results show a minimal latency as low as 0.02 seconds, irrespective of the location, and data store type. Further, integrated APIs provide single notation and are 5% faster than third-party APIs. Our findings thus provide a guide to the proposed integrated RFI system at MeerKAT/SKA radio telescope.