Cloud Computing Resource Scheduling In Hybrid Cloud Environments

Cloud computing has received enormous attention in recent years because of its user self-service provisioning,flexibility,location-free and easy-to-maintain feature,as well as the pay-as-you-go model typically adopted by cloud service providers.With the help of the hybrid cloud model which combines both public and private clouds,corporations and organizations can run applications benefiting from the scalability offered by public clouds while keeping sensitive data inside the fire wall of their own private cloud platforms at the same time.Therefore,hybrid cloud platforms are often the first choice for corporations and organizations to deploy distributed applications.In this paper,we focus on a typical scenario to schedule computing requests in hybrid clouds,trying to minimize the costs incurred while keeping the deadline missing of the requests at an acceptable level.Computing requests submitted can be divided into two categories: some are special requests,and they are specified to be scheduled to the private cloud in consideration of data security,while the other requests,namely normal requests,which are insensitive and can be scheduled to either the private or the public cloud.According to our study,we formulate the scheduling problem in hybrid clouds as an online binary integer programming model originally.However,this scheduling model demands prior knowledge of future request arrivals.By taking advantage of the Lyapunov optimization techniques,we transform it into a one-shot binary integer programming problem which is much easier to solve,and build a scheduling algorithm.We implemented our algorithm and simulation results suggest that our scheduler strikes a good balance between the cost of public clouds and the deadline missing of computing requests.Besides,the proposed scheduler shows a nearly optimal resource utilization rate and a good average scheduling delay.