Flexible Container-Based Computing Platform on Cloud for Scientific Workflows

Cloud computing is expected to be a promising solution for scientific computing. In this paper, we propose a flexible container-based computing platform to run scientific workflows on cloud. We integrate Galaxy, a popular biology workflow system, with four famous container cluster systems. Preliminary evaluation shows that container cluster systems introduce negligible performance overhead for data intensive scientific workflows, meanwhile, they are able to solve tool installation problem, guarantee reproducibility and improve resource utilization. Moreover, we implement four ways of using Docker, the most popular container tool, for our platform. Docker in Docker and Sibling Docker, which run everything within containers, both help scientists easily deploy our platform on any clouds in a few minutes.