Internet routing emulation system and stress testing | IEEE Conference Publication | IEEE Xplore

Internet routing emulation system and stress testing


Abstract:

With the incessant growth of the Internet and the renewal of routing protocols, it becomes a demanding task to test the running characteristics of routing protocol implem...Show More

Abstract:

With the incessant growth of the Internet and the renewal of routing protocols, it becomes a demanding task to test the running characteristics of routing protocol implementations (RPI) in a real large-scale network. We propose a novel approach, named stress testing, to inspect the running characteristics. We compared stress testing with traditional protocol testing, and give the abstract test method. By combining Internet topology generation and RPI, we developed the Internet Routing Emulation System (IRES) as a testbed of stress testing. The architecture and some pivotal issues of IRES are presented. An example of stress testing is given, where IRES is the test bed and a CISCO2600 router is the router under test (RUT). The characteristics of OSPF routing interaction is measured and analyzed between the IRES and RUT. The computation complexity of OSPF RPI in the RUT and the upper bound of the network scale RUT supports are obtained. The experiment shows that stress testing can be widely used to inspect the stability and scalability of other RPI, e.g. RIP, BGP.
Date of Conference: 23 February 2003 - 01 March 2003
Date Added to IEEE Xplore: 02 April 2003
Print ISBN:0-7803-7661-7
Conference Location: Papeete, France

1. Introduction

Routing protocols play a crucial role in computer networks. However, there are few efficient test methods of Routing Protocol Implementations (RPI). For routing algorithms and protocols, researchers often set up a mathematical model to analyze the theoretical characteristics, and then validate the results with simulations [1] [2] [3]. Because computer network systems have the complex characteristics combined with multiple states and randomicity, it is difficult to model them and obtain their characteristics precisely, Furthermore, analyses and simulations cannot measure the characteristics of RPI or implemented systems.

Contact IEEE to Subscribe

References

References is not available for this document.