Headlines

Calendar

Seminar Series

Publications

Media Information

Click here to subscribe for email updates

Netset – A Software Framework for Automation of Network based Tests

Puneet Arora
Date: 2008-12-15
Degree: MS - Computer Networking

Advisory Committee

Dr. Injong Rhee - Committee Chair
Dr. Khaled Harfoush - Committee Member
Dr. Laurie Williams - Committee Member

Abstract

Testing network protocols or networking devices for generating performance benchmarks is integral to computer networking research. The general pattern in conducting such tests is configuring a desired topology over a network test-bed, deploying test software and tools onto network nodes, setting up load traffic on the network, executing the deployed software, gathering data and generating reports for analysis and comparison. A typical example of such a process is performance testing of congestion control protocols. Congestion control protocols are designed to satisfy a complex set of goals and their performance is sensitive to network topology, network delays, and router queue sizes and policies. It is thus non-trivial to construct testing procedures for congestion control protocols. While there exist network emulation test-bed services like Emulab, WanInLab, Deter etc. they limit themselves to merely providing a set of nodes which the developer can use for testing. They provide no means of setting up the test environment, generating test traffic, gathering data and generating reports. As a result developers tend to write their own automation procedures to carry out these steps. Such tendency restricts portability, repeatability and comparison of test procedures and their results. This lack of a generally accepted practice or of a testing tool affects many parts of the networking research community including researchers, students, standardization bodies and developers. To address this issue, we develop a tool for network based testing called Netset. The tool implements a general model which accommodates a wide range of testing processes and provides programming interface to develop specialized tools for a particular category like congestion control protocols. We then apply this tool to two testing scenarios and compare its benefits versus using a manual approach in the same scenarios.

Download the complete document from the NCSU Libraries: download pdf