Dynamic Data Migration Policies for Query-Intensive Distributed Data Environments
Modern large distributed applications, such as telecommunication and banking services, need to respond instantly to a huge number of queries within a short period of time. The data-intensive, query-intensive nature makes it necessary to build these applications in a distributed data environment that involves a number of data servers sharing service load. How data is distributed among the servers has a crucial impact on the system response time. This paper introduces two policies that dynamically migrate data in such an environment as the pattern of queries on data changes, and achieve query load balance. One policy is based on a central controller that periodically collects the query load information on all data servers and regulates data migration across the whole system. The other policy lets individual server dynamically selects a partner to migrate data and balance query load in between. Experimental results show that both policies significantly improve system performance in terms of average query response time and fairness, and communication overhead incurred is marginal.
KeywordsSource Node Destination Node Load Level Average Response Time Server Node
Unable to display preview. Download preview PDF.
- 1.Narendran, B., Rangarajan, S., Yajnik, S.: Data distribution algorithms for load balanced fault-tolerant Webaccess. In: Proceedings of The Sixteenth Symposium on Reliable Distributed Systems, pp. 97–106 (1997)Google Scholar
- 2.Savio, S.: Approximate Algorithms for Document Placement in Distributed Web Servers. IEEE Transactions on Software Engineering, 100–106 (2004)Google Scholar
- 3.Yokota, H., Kanemasa, Y., Miyazaki, J.: Fat-Btree: An Update-Conscious Parallel Directory Structure. In: International Conference on Data Engineering (ICDE), pp. 448–457 (1999)Google Scholar
- 4.Lee, M.-L., Kitsuregawa, M., Ooi, B.-C., Tan, K.-L., Mondal, A.: Towards Self-Tuning Data Placement in Parallel Database Systems. In: International Conference on Management of Data (SIGMOD), pp. 225–236 (2000)Google Scholar
- 6.Watanabe, A., Yokota, H.: Adaptive Lapped Declustering: A Highly Available Data-Placement Method Balancing Access Load and Space Utilization. In: International Conference on Data Engineering (ICDE), pp. 828–839 (2005)Google Scholar
- 7.Feldmann, M., Rissen, J.P.: GSM Network Systems and Overall System Integration. Electrical Communication (1993)Google Scholar
- 9.Wang, T., Yang, B., Gao, J., Yang, D.: Effective data distribution and reallocation strategies for fast query response in distributed query-intensive data environments. In: Zhang, Y., Yu, G., Bertino, E., Xu, G. (eds.) APWeb 2008. LNCS, vol. 4976, pp. 548–559. Springer, Heidelberg (2008)CrossRefGoogle Scholar