Energy-aware online scheduling: Ensuring quality of service for IaaS clouds

This work focuses on the analysis of energy-efficient algorithms for online scheduling in the Infrastructure as a Service (IaaS) type of Cloud computing. The scheduling model is based on service levels that guarantee the quality of service. Each service level is associated to a price per unit of job execution time and a slack factor, which determines the deadline for delivering the results. Once a batch job is submitted to the system, the provider has to decide whether the arriving job can be accepted or must be rejected and processed on external resources. The system maintains the quality of service guarantees for all already accepted jobs. After proposing an original formulation of the problem, we study various schedulers that provide good compromise between income maximization and power consumption minimization. Our experiments and analysis have shown that we can generate schedules with power consumption reduction without degrading the service quality.