Fibre Channel Arbitrated Loop - Performance Analysis of Loop Tenancy Overhead

Abstract

The performance analysis of high-speed storage networks is a challenge due to the inability of most disk based targets to completely utilize a channel. We use a RAM drive to maximize the bandwidth of a 4G fibre channel network and use standard performance metrics to analyze the bottlenecks of the system. We experimentally show the effects of increasing the number of initiators and targets on the performance of an arbitrated loop. We also analyze the impact of increasing the length of the loop beyond the standard. Finally, we conclude how the performance of a device on the loop can plummet when there is another device with higher priority not using fairness.