Abstract:

Data networks are critical to leverage an increasingly global and connected world. As such, higher education degrees such as telecommunications or computer science/engineering, commonly incorporate courses addressing networking technologies.

Traditionally, laboratory classes in this field make use of real equipment such as, personal computers, switches, routers, firewalls, access points and other physical hardware to deploy networking scenarios for hands-on experience. The main plus of this learning model is to familiarize students with equipment used in real-life production environments.

This paradigm, however, has some major drawbacks, namely:
• substantial initial investment is required to acquire networking equipment;
• scenario setup and configuration can consume a large portion of class time;
• scalability is limited if the number of enrolled students grows;
• limited laboratory access outside regular class hours (time barrier);
• students cannot access equipment remotely which means that physical presence is necessary (geographic barrier);
• continuous investment is required to keep up with technology updates;

In order to overcome these drawbacks, this paper proposes AMBIVeRT, a platform that replaces the traditional network laboratory with a virtual class environment, emulating typical networking scenarios addressed in computer network related courses. The platform’s architecture is based on several virtual machines (VM) connected through virtual networks, which implement two different domains: an administration domain for configuration and initial setups, and a student domain, used by scholars to configure and explore different networking scenarios. The platform has been deployed in one single server (Cisco UCS C200 M2 High-Density Rack Server) running VMware vSphere Hypervisor.

Each student/group has access to an individual virtual workbench, consisting of three different VMs connected in the same VLAN. The first VM houses a GNS3 VM which functions as a back-end headless network emulator. The other two VMs run commonly used operating systems, such as Windows Server and Linux, and act as the frontend to the network emulator. Therefore, the platform supports multiple isolated virtual workbenches.

Several network scenarios have been tested, from simple network architectures with a few virtual equipment’s to complex CCIE scenarios, with 18 virtual appliances, some emulating advanced networking devices.

This paper describes the AMBIVeRT platform’s architecture, functionalities and current limitations. System benchmark is also presented, with special focus on comparing the performance for different scenario’s complexity and number of active workbenches. Results have shown that the AMBIVeRT platform is highly scalable and a serious alternative as a replacement for physical networking laboratories both for local and remote teaching.

Moreover, the paper compares this platform with traditional physical networking laboratories and with other virtualized network environments, concerning available features, costs and scalability.

Keywords:

Virtualization, b-learning, education, computer networks, emulation.