Imaging SKA-Scale Data on the Cloud, a Supercomputer, and an In-House Cluster

We have investigated how one would manage to process SKA scale data volumes using three very different computing environments: a moderate sized cluster, such as a group like ICRAR hosts and controls; a high performance computing cluster that would be provided by a national facility such as the Pawsey centre; and a cloud computing environment, such as provided by the Amazon Web Services. This allowed us to explore three very different approaches, all of which would be of the scale accessible to groups, such as ours, via in-house capital expenditure, via competitive applications for resources on national infrastructure or via cumulative operational expenditure.