作者: Gandhimathi Velusamy , Ricardo Lent
关键词:
摘要: The low maintenance requirement, capacity scalability, and pay-as-you-go properties of cloud computing are attractive for the virtualized deployment diverse web services. Web traffic is typically handled by multiple server mirrors that spatially dispersed to satisfy expectations a large number worldwide users. Since energy consumption each depends on its workload, use routing opens possibility reducing operational costs through exploitation regional temporal differences in pricing at mirroring sites. On downside, shared nature network brings potential latency issues could impact quality service many applications. In this paper, we report experimental results obtained from system uses learning automata, reinforcement approach make dynamic decisions based cost quality-of-service criteria cloud. experiments were conducted using 24 nodes running CloudLab with time-varying prices modeled real data.