Software defined federated cyber-infrastructure for science and engineering

Scientific applications are moving towards dynamic and data-driven workflows with changing resource requirements. Oftentimes, these workflows combine dynamically changing resource requirements and very large computational and throughput demands. Federated computing has been explored in various contexts and has been demonstrated to be an attractive and viable model for effectively harnessing the power offered by distributed resources. However, resources themselves may also exhibit dynamic behavior with varying capabilities, capacities, and availability over time. Hence, being able to provision an appropriate computational environment at the right time is a challenging issue that cannot be solved using classic federation models where a user is presented with a fixed set of resources. In this paper we argue that an elastic execution infrastructure based on the dynamic federation of resources can help with varying application requirements. In particular, we propose a federation architecture that combines cloud abstractions with ideas from software-defined environments to offer users an integral solution for solving large scale problems in dynamically federated cyber-infrastructure.