How can large amounts of data be processed without investing in any Big Data processing infrastructure and only paying for the amount of time the processing resources are actually used?