摘要
为解决目前集群文件服务系统中负载均衡算法的缺陷,充分考虑文件本身存在的差异,降低大文件分配到负载重的服务器上而造成负载失衡的可能性,避免频繁收集节点的负载信息造成大量额外的网络开销而降低系统吞吐量,提出了改进的负载均衡算法。该算法综合考虑文件负载量、服务器的硬件配置及其实时负载量,引入对上传文件大小和数量的计算,改变服务器负载量定时更新为定量更新。实验结果表明,该算法减少了客户的等待响应时间,提高了系统吞吐量,达到了良好的负载均衡。
At present in the cluster file service system, the existing load balancing algorithm has not fully considered the diffe- rences among the file itself, which leads to large files may be assigned to a heavy load on the server and cause load imbalance. And frequently collecting load information cause additional network overhead that lower system throughput. The improved algo rithm introduces that considering the file size, server hardware configuration and real-time load, the upload file size and number are calculated, and its regularly updated server load is changed to quantitative updated. The experimental results show this algo rithm reduces the response time and improves the capacity of throughput, and achieve a good load balancing effect.
出处
《计算机工程与设计》
CSCD
北大核心
2013年第8期2754-2758,共5页
Computer Engineering and Design
基金
天津市应用基础与前沿技术研究计划重点基金项目(12JCZDJC21200)
关键词
文件服务系统
负载均衡
集群
文件负载
服务器负载
distributed file service system
load balancing~ cluster
file load
server load