摘要
在分析基于iSCSI协议的IP存储广域网(IP-SWAN)的数据传输流程的基础上,提出了IP-SWAN系统的随机Petri网模型和一种新的基于变迁的串并联等价变换计算系统I/O响应时间的SPNA方法。基于SPNA方法的性能分析和预测结果表明,在低速网环境下,网络传输速率是系统性能的主要瓶颈,提高网络带宽可以有效地改进整个系统性能。在高速网环境下,提高中心节点的读缓存命中率和CPU处理效率,可以降低系统的I/O响应时间。
Through analyzing the data flow in iSCSI-based IP-SWAN system, this paper proposes stochastic Petri net model (SPNM) of IP-SWAN system and a novel SPNA method for computing system I/O response time using equivalent conversion of serial or parallel transition. The performance evaluation and prediction are made by SPNA method. Theoretical analysis shows that bandwidth is the bottleneck of IP-SWAN performance in low-speed network, the I/O response time is reduced through enhancing the cache hit rate for read operation and CPU processing power at central server in high-speed network.
出处
《计算机工程》
CAS
CSCD
北大核心
2007年第17期114-116,共3页
Computer Engineering
基金
国家自然科学基金(90612001)
天津市高等学校科技发展基金(20041603)