On Optimizing MMVEs in Network-Aware Clouds

Network operators will soon cooperate with traditional cloud providers to offer network-virtualization-based converged cloud services, which are referred to as network-aware clouds. Network-aware clouds allow network operators to share income with Over-The-Top (OTT) providers by providing them with end-to-end network QoS guarantees. For MMVE providers, leveraging the computation, storage, and communication resources offered by network-aware clouds for the best MMVE QoE levels is crucial to their success. In this paper, we point out a main research challenge: optimally placing various fine-grained MMVE tasks across heterogeneous clouds, which provide diverse computation and storage QoS guarantees (in data centers) and communication QoS guarantees (end-to-end). Via real experiments, we demonstrate the potential of network-aware clouds on improving the QoE of MMVEs. Achieving the optimal QoE level, however, is no easy task because of the dynamic nature of networks and virtual environments and the complex interplay between cloud QoS guarantees and MMVE QoE metrics, such as responsiveness, precision, and fairness. Throughly addressing the task placement problem is our current work.