Network storage system simulation and performance optimization

Abstract

The simulation modeling has been proven to be an effective approach for system performance analysis and new algorithm evaluation. However, there are few simulation tools that can support detail enough simulation. In order to accurately conduct performance analysis on network storage system, a network storage system simulation and modeling tool that can simulate the network storage system up to Frame level is developed in this thesis. The simulation tool is calibrated and validated through experiments. The comparison between actual experiments and simulation results shows that the simulation model can achieve high accuracy as to 3% mismatching for read I/Os. A new command schedule algorithm is proposed to optimize the overall storage system I/O performance. The simulation results are analyzed and the measurements show that the performance gains achieved by the algorithm are up to 50% improvement comparing to the normal schedule in certain circumstances.