Bruce Berriman (IPAC) and Ewa Deelman (Information Sciences Institute, University of Southern California): How To Use Cloud Computing To Do Astronomy

Cloud computing has been receiving a lot of publicity lately as a new platform for scientific computing and data storage. We give a practical guide for astronomers on how to use cloud computing to support data processing and storage, especially for data driven workflow applications. The pay-as-you-go business model used by cloud providers require that users become aware of how their jobs use resources. We will describe the kinds of applications that are best run on cloud platforms, and show how costs depend critically on how applications use memory, disk space and processing cycles. We show how performance on cloud platforms compares with that on other types of platform. Cloud providers generally make compute resources available, but little support for systems administration and configuration, and for provisioning resources to act as a cluster to run compute-intensive applications. We describe tools that make this process easy for even an inexperienced user.