Multi-criteria scheduling of Bag-of-Tasks applications on heterogeneous interlinked clouds with simulated annealing

Abstract

Cloud computing has spurred the creation of a multitude of services that use the cloud to deliver their products on-demand. Behind it, stand multiple “Cloud Providers” that in the past few years have created data-centers, spread around the world, creating a mesh of distributed resources that can meet high availability and quality of service requirements. The growing number of cloud clients demand reliability, performance and better cost-to-performance ratios. Recently, scientific research has focused on the optimization of interlinked cloud systems, an aim which requires strategies for allocation of resources and distribution of computing tasks between them, while also considering their cost along with any factors that may differentiate them. In this study, we have evaluated the use of simulated annealing and thermodynamic simulated annealing in the scheduling of a dynamic multi-cloud system with virtual machines of heterogeneous performance serving Bag-of-Tasks applications. The scheduling heuristics applied, consider multiple criteria when scheduling said applications and try to optimize both for performance and cost, while also taking into account the heterogeneity of the virtual machines. Simulation results indicate that the use of these heuristics can have a significant impact in performance while maintaining a good cost-performance trade-off.