论文部分内容阅读
测试负载生成是网络性能测试的重要环节之一。为了在测试过程中模拟复杂的数据流量环境,常常通过组合大量特征各异的流来近似实际网络流量。流的重要属性是其速率与突发程度,文中将此定义为流模式。在使用多个流来模拟复杂的网络环境时,一个重要问题是如何调度多个流,以保证各个流的流模式。提出了一种调度方法GRR(groupingroundrobin),将所有流分成若干小组后分别在组内和组间两级轮询调度,可在较低的时间和空间复杂度内实现调度目标。算法已在一个高速网络性能测试系统IP-TEST中得到应用,达到了1Gb/s的线速发送速率,并且支持突发流模式。
Test load generation is an important part of network performance testing. In order to simulate a complex data traffic environment during testing, the actual network traffic is often approximated by combining a large number of distinct flows. The important attribute of a stream is its rate and burst level, which is defined as a stream mode in the text. When using multiple streams to simulate a complex network environment, one important issue is how to schedule multiple streams to ensure the flow pattern of each stream. A scheduling method, called GRG (groupingroundrobin), is proposed. After all the flows are divided into several groups, they are scheduled round-robin at the intra-group and inter-group levels respectively, which can achieve the scheduling goals with lower time and space complexity. The algorithm has been applied in a high-speed network performance test system IP-TEST, achieving a wire-speed transmission rate of 1Gb / s, and supports burst mode.