Revenue-sensitive scheduling of multi-application tasks in software-defined cloud

The development of cloud computing attracts a growing number of corporations to implement their applications in data centers. The increase in variety and amount of applications in data centers that support software-defined networking (SDN) protocols makes it a big challenge to maximize revenue for data center providers. However, current SDN controllers just consider latency optimization in network and do not consider latency in virtual machines (VMs), and therefore revenue loss may occur. Different from current studies, this work aims to maximize revenue of a software-defined cloud provider. A Revenue-sensitive Scheduling of Multi-application Tasks (RSMT) method is then proposed to increase the revenue of a cloud provider. It is realized by jointly determining optimal routing paths and VMs for multi-application tasks. Simulation based on real-life task data demonstrates that compared with several current algorithms, RSMT can produce the efficient schedules that increase the cloud provider's revenue and decrease round trip time of multi-application tasks.