The packet pair mechanism has been shown to be a reliable method to measure the bottleneck link bandwidth of a network path. However, the use of packet pairs to measure available bandwidth has had more mixed results. In this paper we study how packet pairs and packet trains can be used to estimate the available bandwidth on a network path. As a starting point for our study, we construct the gap model, a simple model that captures the relationship between the competing traffic and the input and output packet pair gap for a single hop network. We validate the model using measurements on a testbed. The gap model shows that the initial probing gap is a critical parameter when using packet pairs to estimate available bandwidth. Based on this insight, we propose a new technique to measure the available bandwidth -- the Initial Gap Increasing (IGI) algorithm, which experimentally determines the best initial gap for measuring available bandwidth. Our experiments show that measurements that take 4-6 round trip times allow us to estimate the available bandwidth to within about 10%.

27 pages

Return to: SCS Technical Report Collection
School of Computer Science homepage

This page maintained by reports@cs.cmu.edu