Experiment-Driven Modeling of Open-Source Internet Traffic Generators

Abstract

This paper is a first step toward the goal of evaluating the measurement accuracy of the interdeparture times (IDTs) provided by software traffic generators. To provide evidences of the proposed methodology, in this paper, we focus our attention on the experimental characterization of the internal generation cycle of a well-known, open source generator, namely, distributed internet traffic generator. We tested the traffic generator for different systems, and under the best possible conditions, i.e., with the minimum system load. The resulting performance may be seen as the ideal limit the generators can tend to. In this paper, we present the rationale for the activity, the underlying methodology and a set of tests aimed at highlighting the relevance of the clock resolution in the accuracy of IDT. In addition, we present a statistical model for the IDT of packets. The model allows to capture the effects of the various phenomena that affect this important generation parameter. We have also performed a set of experiments with another traffic generator to show the generalizability of the results. These results show that the contributions proposed in this paper are general and valid for a wide set of software-based traffic generators

How to Quantify Trust in Your Network Emulator?

“…Due to their cheapness and high degree of flexibility and versatility, the second ones are very widespread in practice [11] and, mainly thanks to the open-source feature, their employment is even more increasing also in the field of research. However, since they are based on general purpose platforms, their working strictly depends on the interaction of hardware and software sections of the hosting platform and their performance cannot be a priori guaranteed, as also proved in [12] - [15]. Therefore, prior to use such systems, a performance characterization and validation should be made for certifying the ability of these software to accurately reproduce the wanted network conditions.…”

“…In this framework, focusing the attention on GPNE, starting from the past experiences in metrological characterization of computer networks devices and services [9], [15] - [18], a methodological approach is proposed to measure and analyze the performance of such systems. In a more detail, a suitable measurement setup is designed and characterized for analyzing the performance of GPNE.…”

Tools for Network Traffic Generation - A Quantitative Comparison

“…Network Traffic Generators (NTGs) are vital tools in the networking and security fields [1,2]. They mainly focus on creating and injecting crafted network traffic into a network that can later be consumed by other devices in a controlled fashion.…”

“…They mainly focus on creating and injecting crafted network traffic into a network that can later be consumed by other devices in a controlled fashion. This is usually done to assess the performance of networks, infrastructures, security controls, and other aspects [1]. They can be implemented over both hardware and software platforms.…”

“…Further, this kind of traffic generation tools is generally developed by firms and preconfigured to carry out a specific type of tests, and therefore difficult to customize [2]. Whereas, software-based generators are usually open-source, cheaper, but slower and less accurate [1,3]. These tools are the most wildly used in the networking field mainly due to their high flexibility and open-source nature that allows for easy modifications and extensions based on specific research goals [1].…”