A method for inexpensively simulating global Internet protocol wide-area networks in a laboratory environment

With the widespread adoption of Internet protocol (IP) as the de facto data communications protocol of the commercial, government, and military communities, more and more staff at all levels have become involved in the development, engineering, deployment, and operation of IP-based networks and the applications that ride over them. Unfortunately, the complexity of these systems precludes simply "doing it", as key technologies and components must be extensively simulated and tested in off-line laboratory environments prior to operational deployments. To support this testing, it is necessary to simulate the complete environment that systems will be deployed into, including Local- and Wide-Area Networks (LANs and WANs) and the links that comprise them. Unfortunately, deploying full-scale WANs in a laboratory environment can rapidly become cost-prohibitive. A single backbone-level switch or router can easily cost hundreds of thousands of dollars, and sites with complex WAN connectivity such as redundant links with dynamic routing can require connections to multiple backbone-level switches and routers. Additional complexity ensues when lab engineers must reproduce or simulate WAN links with accurate bandwidth, latency, and reliability characteristics. All of these considerations can translate into multi-million dollar lab infrastructures to support the simplest application and system development projects. This is at the same time that accurate lab testing is necessary not just at the prime contractor level, but also at the sub-contractor, engineer, and even operator levels. In short, the cost and complexity of creating accurate lab WAN environments is growing at the same time that there is increasing need for accurate and cost-effective lab-based WANs at the lowest development and operational levels. This paper looks at the considerations associated with accurately simulating WANs in a lab, including bandwidth, latency, reliability, redundancy, and complexity issues. It examines workarounds that are commonly used by engineers to manage the cost of lab WAN environments, and some of the dangers and pitfalls associated with those techniques. Finally, it proposes a method for effectively simulating global WANs using inexpensive off-the-shelf components in such a way that maximizes flexibility, functionality, and testing integrity while minimizing the overall lab infrastructure cost.

INTRODUCTION

When developing modern computer-based communications systems, a large number of factors must be considered and tested against. This testing requires realistic lab environments that accurately reproduce the vagaries of the real world. Only through extensive testing can modern network-based systems be properly validated for operation. The remainder of this paper describes an approach for easily simulating large Wide Area Networks (WANs) in a laboratory environment. Fig. 1 Typical Application / System for Testing