In this paper, we present a low cost and flexible testbed to evaluate the performance of available bandwidth estimation tools in a common and controlled environment. In addition, we include a model based on a network of M/M/1 queues to have an analytical reference to compare the experiments with. We then utilize the proposed testbed and model to evaluate the performance of Pathload, IGI, and Spruce, three well-known bandwidth estimation tools. Our main results indicate that Pathload is the most accurate tool but the slowest to converge. IGI, on the other hand, is the fastest tool but the least accurate. Spruce is the least intrusive tool with intermediate accuracy and convergence time.