FRLDM: A Self-Optimizing Algorithm for Data Migration in Distributed Storage Systems
作者： Tao Wang , Shihong Yao , Zhengquan Xu , Shan Jia , Lizhi Xiong
DOI: 10.2991/IWMECS-15.2015.114
关键词: Data migration 、 Distributed computing 、 Computer science 、 Distributed data store 、 Service (systems architecture) 、 Response time 、 Heuristic 、 State (computer science) 、 Reinforcement learning 、 Fuzzy reinforcement learning
摘要: In distributed storage systems, data migration is an efficient method for improving system resource utility and service capacity, balancing the load. However, user accessing changing over time state of a in unpredictable stochastic fluctuation, hence traditional heuristic policy- based methods are hard to work such environment. This paper proposes fuzzy reinforcement learning online named FRLDM which can enable systems self-optimize dynamically choose candidate on their recent access pattern current system, thus minimizing average response time. The experimental results prove that improve accesses performance significantly compared with policy-based methods.
sci-hub.st 参考文章(14)
Kary Främling, Replacing eligibility trace for action-value learning with function approximation. the european symposium on artificial neural networks. pp. 313- 318 ,(2007)
Tao Wang, Shihong Yao, Zhengquan Xu, Shan Jia, DCCP: an effective data placement strategy for data-intensive computations in distributed cloud computing systems The Journal of Supercomputing. ,vol. 72, pp. 2537- 2564 ,(2016) , 10.1007/S11227-015-1511-Z
Siegmund Duell, Steffen Udluft, Volkmar Sterzing, Solving Partially Observable Reinforcement Learning Problems with Recurrent Neural Networks Neural Networks: Tricks of the Trade (2nd ed.). pp. 709- 733 ,(2012) , 10.1007/978-3-642-35289-8_38
Michael L. Littman, Reinforcement learning improves behaviour from evaluative feedback Nature. ,vol. 521, pp. 445- 451 ,(2015) , 10.1038/NATURE14540
Gong Zhang, Lawrence Chiu, Ling Liu, Adaptive Data Migration in Multi-tiered Storage Based Cloud Environment international conference on cloud computing. pp. 148- 155 ,(2010) , 10.1109/CLOUD.2010.60
Qi Zhou, Peng Shi, Shengyuan Xu, Hongyi Li, Adaptive Output Feedback Control for Nonlinear Time-Delay Systems by Fuzzy Approximation Approach IEEE Transactions on Fuzzy Systems. ,vol. 21, pp. 301- 313 ,(2013) , 10.1109/TFUZZ.2012.2213259
Jun Wu, Xin Xu, Pengcheng Zhang, Chunming Liu, A novel multi-agent reinforcement learning approach for job scheduling in Grid computing Future Generation Computer Systems. ,vol. 27, pp. 430- 439 ,(2011) , 10.1016/J.FUTURE.2010.10.009
Yanjie Li, Baoqun Yin, Hongsheng Xi, Finding optimal memoryless policies of POMDPs under the expected average reward criterion European Journal of Operational Research. ,vol. 211, pp. 556- 567 ,(2011) , 10.1016/J.EJOR.2010.12.014
Zhipeng Tan, Wei Zhou, Dan Feng, Wenhua Zhang, ALDM: Adaptive Loading Data Migration in Distributed File Systems IEEE Transactions on Magnetics. ,vol. 49, pp. 2645- 2652 ,(2013) , 10.1109/TMAG.2013.2251616
Bin Dong, Xiuqiao Li, Limin Xiao, Li Ruan, Binbin Yu, Self-Acting Load Balancing with Parallel Sub File Migration for Parallel File System computational sciences and optimization. ,vol. 2, pp. 317- 321 ,(2010) , 10.1109/CSO.2010.37