A SLA-based method for big-data transfers with multi-criteria optimization constraints for IaaS

When one pays for a Cloud Service he wants the Service to be compliant with standards and to respect his needs (respect the Service Level Agreement). When manipulating big-data (like picture collections, satellite images or digital libraries), transfers must be optimized. The Cloud Infrastructure itself utilizes big-data transfers while migrating Virtual Machines inside an Infrastructure as a Service system. In this paper we discuss data transfers in the cloud, which affects performance in the case of Virtual Machine migration and of user submitted bigdata transfers, by moving resources where they are needed. We suggest a scheduling policy and offer two greedy scheduling algorithms that minimize individual transfer times. This is mostly important for being able to have a High Performance Computing environment inside the cloud, for scenarios in which individual data transfer times need to be but when the data is transferred is not an issue. We present an empirical validation based on simulation experiments.