Details

Autor(en) / Beteiligte

• Zhong, Jiang
• Chen, Hao
• Yang, Lei
• Lu, Zeguang
• Yin, Guisheng
• Sun, Guanglu
• Jing, Weipeng
• Chen, Wen'guang
• Han, Qilong
• Zhao, Gansen
• Chen, Wenguang

Titel

Energy Conservation Strategy for Big News Data on HDFS

Ist Teil von

• Big Data Technology and Applications, 2016, Vol.590, p.59-73

Ort / Verlag

Singapore: Springer Singapore Pte. Limited

Erscheinungsjahr

2016

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• In this paper, an energy-conservation Hadoop Distributed File System (HDFS) oriented to massive news data is proposed based on news access pattern, in order to reduce energy consumption of big news data storage system. First, divide all data nodes into real-time responding hot data nodes and standby cold data nodes. To make a good balance between data access performance and energy-conservation, this paper takes two strategies of priority allocation, named Active State Node Priority (ASNP) and Lower Than Average Utilization Rate Node Priority (LANP), to mostly guarantee the balance of data distribution in cluster in order to obtain a good data access performance. It also confirms the opportunities to move data from hot data nodes to cold data nodes is based on the access pattern of news data and develops a simulating experimental platform that can evaluate energy consumption of any file accessing operation under any different storage strategies and parameters. Simulation experiments shows that strategies proposed in this paper saves 20 %–35 % energy than traditional HDFS and 99.9 % responding time of reading files will not be affected, with an average of 0.008 %–0.036 % time delay.