摘要
Hadoop云存储架构的设计初衷是实现大文件的高效存储处理,但在处理移动终端下诸如图片等小文件时会引起名称节点索引提取速度过慢和数据节点存储空间利用率不高等问题。针对这一问题,提出一种小文件归档的方案FHAR。方案综合考虑移动终端用户访问的实时性、名称节点服务器内存使用率、数据节点存储空间利用率等方面,利用双层索引的归档技术结合FAHP(模糊多属性决策理论)的系统负载预测算法实现系统的负载均衡,提高服务效率。同时利用数据预取机制对访问操作进行优化。仿真结果表明,该方案有效提高了节点的存储效率与用户访问的实时体验性。
Hadoop is a cloud storage architecture which was originally designed to realize the effi- cient storage of the large file processing, however, it will cause a low extraction speed of the name node index and low utilization of the data node storage space when processing the small file in the mobile terminal such as pictures. To solve this problem, we proposed a approach FHAR Considering the time of access to mobile users, the usage of name node memory and the efficiency of data stor- age. Combined with the theory of FAHP（ Fuzzy Analytical Hierarchy Process） prediction algorithm and making with double indexing archiving technique to achieve load balancing of the system load and improve the efficiency of s.ervice. At the same time use data prefetching mechanism for accessing operation optimization. The simulation results show that this approach can effectively improve the u- tilization of the nodes and the real - time access.
出处
《无线通信技术》
2014年第1期32-36,共5页
Wireless Communication Technology
