基于并发跳表的云数据处理双层索引架构研究.PDF

DOI ：１０ ．７５４４/issn１０００‐１２３９ ．２０１５ Journal of Computer Research and Development ５２（７）：１５３１ １５４５ ，２０１５ 　 　 　 　 　 　 （　 　 ６５００９１） （zw ei＠ y nu ．edu ．cn ） Concurrent Skiplist Based Double‐Layer Index Framework for Cloud Data Processing Zhou Wei ，Lu Jin ，Zhou Keren ，Wang Shipu ，and Yao Shaow en （School of Sof tw are ，Y unnan Univ e rs ity ， K unming ６５００９１） Abstract 　 Cloud data processing play s an essential inf rastructure in cloud sy stem s ．Without efficient structures ，cloud sy stem s canno t support the necessary high throughput and provide services f or millions of users ． How ever ，mo st existing cloud storage sy stems generally adopt a distributed Hash table （D HT ） approach to index data ， w hich lacks to support range‐ uery and dy namic real‐time q character ．It is necessary to generate a scalable ，dy namical and multi‐ uery f unctional index structure q in cloud environment ．Based on the summary and analy sis of the double‐lay er index sy stem s f or cloud storage ，this paper provides a novel concurrent skiplist based double ‐lay er index （ref erred as CSD‐ index ） f or cloud data processing ． T w o‐lay er architecture ， w hich can breakthrough single machine memory and hard drive limitation ，is used to ex tend index ing scope ． Online migration algorithm of skiplist ’s nodes betw een local servers is used to make dy namic load‐balancing ． T he details of the design and the implement of the concurrent skiplist are discussed in this paper ．Optimistic concurrency control （OCC ） technique is introduced to enhance the concurrency ．T hrough concurrent skiplist CSD‐ index improves the load bearing capacity of the global