Resource management is one significant issue of Cloud Computing in both academic research and commercial applications.Well-utilized architectures help system allocate resources effectively and improveQoS of Cloud Clients.Traditionally,one-way scheduler is implemented based on client requests,which leads obvious lagging deviationdefects.To solve the problems,the proposedsystem architecture aiming at eliminating lagged effect and wastage of resources is constructed based on closed-loop control theory.Customer requests and resource provisions are presented as input and output respectively.With coupling functions of feedback control,the schedulers constantly adjust supply and demand dynamically in order to achieve equilibrium state ultimately.Multiple schedulers are consolidated and plugged into the framework with control mechanism to regulate resources balance and degrade workload fluctuation while guaranteeing SLA.Simulations results obtained contrastively justify our expectation on the approach,which is practical and effective,and capable of achieving the objective.
展开▼