Abstract:
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.
Key words:
IP-SWAN,
performance evaluation,
stochastic Petri net
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。
关键词:
IP存储广域网,
性能评价,
随机Petri网
CLC Number:
LIU Jun; CUI Bao-jiang; XIE Guang-jun; LIU Jing. Performance Evaluation of IP-SWAN System Based on Stochastic Petri Net[J]. Computer Engineering, 2007, 33(17): 114-116.
刘 军;崔宝江;谢广军;刘 璟. 基于随机Petri网的IP-SWAN系统性能评价[J]. 计算机工程, 2007, 33(17): 114-116.