VECIMS 2003 - International Symposium on Virtual Environments, Human-Computer Interfaces, and Measurement Systems Lugano, Switzerland, 27-29 July 2003
Performance Comparison of Different Techniques for Available Bandwidth Measurement in Packet Switched Networks Leopoldo Angrisani(1), Salvatore D’Antonio(2), Michele Vadursi(3), Giorgio Ventre(1) (1) Dipartimento di Informatica e Sistemistica, Università di Napoli ‘Federico II’, Via Claudio 21, 80125 Napoli, Italy Phone: +39-081-7683170, Fax: +39-081-2396897, Email:
[email protected],
[email protected] (2) CINI-Consorzio Interuniversitario Nazionale per l’Informatica, Via Diocleziano 328, 80125 Napoli, Italy Phone: +39-081-7682894, Fax: +39-081-7682950, Email:
[email protected] (3) Dipartimento di Ingegneria Elettrica, Università di Napoli ‘Federico II’, Via Claudio 21, 80125 Napoli, Italy Phone: +39-081-7683866, Fax: +39-081-2396897, Email:
[email protected]
Abstract – As variety of applications and services that can be provided on packet switched networks has strongly increased over the last years and keeps on increasing, the need for an accurate network monitoring has become an important challenge to be dealt with accurately. Among the different parameters that can be considered for the traffic monitoring on a path, this paper focuses the attention on the available bandwidth. Three different techniques and related software implementations aimed at measuring available bandwidth are taken into consideration. The purpose of this work is to provide a performance comparison of the three techniques under different operating conditions. To this aim, a proper measurement station has been set up by the authors.
I. INTRODUCTION Traffic monitoring provides network operators with a detailed view of the state of their networks. In particular, congestions can be detected through periodic summaries of traffic load and packet loss on individual links; parts of the network exhibiting high delay or loss and routing anomalies, such as forwarding loops, can be identified by means of active probes between pairs of points in the network. Network operators can exploit results provided by the traffic measurement activity in order to perform fundamental activities: (i) usage-based accounting, (ii) traffic profiling, (iii) traffic engineering and (iv) attack/intrusion detection. Among the different parameters peculiar to traffic monitoring, one of the most significant is the available bandwidth of a path, because of its important role in traffic engineering algorithms, on which bandwidth allocation and choice of the best connection path are based. Different techniques and related software tools, aimed at measuring available bandwidth, have been presented in the literature. This paper takes into consideration three of them: (i) Pathload [1], which sends packet streams on the path, and tunes the stream rate after analyzing the trend of one-way delays (OWDs), until an upper and a lower bound for the available bandwidth are found; (ii) IGI [2], which sends packet trains, characterized by increasing gaps between two adjacent packets and then evaluates the available bandwidth on the basis of the variation of the gaps at destination; (iii) Iperf [3], which sets up a TCP (Transmission Control
Protocol) connection and measures the available bandwidth as the ratio between the number of bytes sent over a determined time interval and the time interval itself. Although the related authors have already carried out some experimental tests, an exhaustive and reliable metrological characterization of these techniques has not been executed yet. In the absence of such a characterization, an accurate performance comparison cannot be gained. This paper aims at overcoming this absence. Specifically, a proper comparison of the performance of the aforementioned techniques, for a specified network topology and in terms of measurement compatibility, repeatability, and bias, is drawn. Advantages and limits of these techniques are also going to be highlighted, thus representing the basis for possible future improvements. II. TECHNIQUES FOR AVAILABLE BANDWIDTH MEASUREMENT: A BRIEF OVERVIEW Before giving some details concerning the measurement techniques analyzed and related software implementations, the definition of available bandwidth is briefly outlined. On a network path, constituted of N link, L1,…,LN, each of which can transfer data at a rate Ci, i=1,…,N, it is possible to define: (i) the bottleneck link bandwidth (or, equivalently, the path capacity) as Cb=min(C1,C2,...,CN) and (ii) the available bandwidth as A=min[C1(1-u1), C2(1-u2), …, CN(1-uN)], where u1,…,uN are the percentages of usage of the links. As stated before, three techniques have been taken into account: Pathload [1], IGI [2], and Iperf [3]. Pathload consists of two components: the process called SND, which runs at sender, and the process called RCV, which runs at the receiver. The process SND transmits a periodic UDP (User Datagram Protocol) packet stream to RCV. Let us suppose that the transmission rate of the stream is R bits per second. Since SND timestamps each packet prior to its transmission with a timestamp ti, RCV can compute the OWD of each packet. Upon the reception of the entire stream, RCV inspects the sequence of OWDs in order to check if the transmission rate R is larger than the available bandwidth A. Indeed, when R>A, the OWDs of the stream packets are expected to have an increasing trend. On the contrary, when R
