Dynamic replica creation strategy based on file heat and node load in hybrid cloud

Replica creation strategy is one of the important research directions of the distributed file system in the hybrid cloud environment. However, traditional replica creation strategy just simply calculated the file heat based on the number of accesses to the file within a period of time. Besides, creating too many copies will seriously affect the performance of the node without considering the node load. In order to solve this problem, the improved dynamic replica creation strategy based on file heat and node load is presented in this paper combined with the characteristics of the hybrid cloud environment. File heat of history and current access frequency of three cycles and change rate of file are considered comprehensively in the calculation of the heat based on LRFU(Least Recently Frequently Used). Combined with the node load, the average heat and the average load are used to adjust the number of copies in this paper, which can adapt to the changes of the environment dynamically. Experiments show that with changes of file access and traffic intensity, the improved strategy is sensitive to access frequency, which can adaptively adjust the number of copies, reduce the average response time, and achieve better load balance of cluster.

[1]  Haifeng Chen,et al.  Proactive Workload Management in Hybrid Cloud Computing , 2014, IEEE Transactions on Network and Service Management.

[2]  Hong Zhao,et al.  The Model of Data Replica Adjust to the Need Based on HDFS Cluster , 2012, 2012 Fifth International Conference on Business Intelligence and Financial Engineering.

[3]  Gholamhossein Dastghaibyfard,et al.  A dynamic replica management strategy in data grid , 2012, J. Netw. Comput. Appl..

[4]  Hairong Kuang,et al.  The Hadoop Distributed File System , 2010, 2010 IEEE 26th Symposium on Mass Storage Systems and Technologies (MSST).

[5]  Xiaorong Li,et al.  Multi-Objective Game Theoretic Schedulingof Bag-of-Tasks Workflows on Hybrid Clouds , 2014, IEEE Transactions on Cloud Computing.

[6]  Yongwei Wu,et al.  µLibCloud: Providing High Available and Uniform Accessing to Multiple Cloud Storages , 2012, 2012 ACM/IEEE 13th International Conference on Grid Computing.

[7]  Huang Mengxing,et al.  A Strategy of Dynamic Replica Creation in Cloud Storage , 2013, CloudCom 2013.

[8]  Kyungbaek Kim,et al.  Time-Related Replication for P2P Storage System , 2008, Seventh International Conference on Networking (icn 2008).

[9]  Eric Anderson,et al.  Capture, Conversion, and Analysis of an Intense NFS Workload , 2009, FAST.

[10]  R. Sepahvand,et al.  Replication and scheduling Methods Based on Prediction in Data Grid , 2011 .

[11]  Qiongxin Liu,et al.  Dynamic Data Replication Based on Access Cost in Distributed Systems , 2009, 2009 Fourth International Conference on Computer Sciences and Convergence Information Technology.