Teaching Computer Networks with GENI Fraida Fund NYU Tandon School of Engineering ffund@nyu.edu Shivendra S. Panwar NYU Tandon School of Engineering panwar@nyu.edu ABSTRACT A popular course in computer networks at NYU Tandon involves several hours each week of small-group in-person lab work. Follow- ing NYU’s move to online learning in response to COVID-19, we transitioned a class of 150 students from this in-person lab setting to an equivalent online experience using the GENI testbed, an open infrastructure for networking and distributed systems research and education. In this white paper, we describe the publicly available resources that enabled this transition, our experiences teaching a computer networking lab course on GENI, and outstanding chal- lenges that remain for future semesters. 1 INTRODUCTION The move to online learning in response to COVID-19 has been especially challenging for in-person lab-based courses in computer networks. In the Electrical and Computer Engineering department at NYU, a popular course in computer networks is aimed at grad- uate students who may not have had hands-on experience with networks in their undergraduate education. The course follows the textbook TCP-IP Essentials: A Lab-Based Approach [5], and fea- tures several hours per week of small-group in-person lab activities, where students work together to carry out a series of fundamen- tal experiments on topics in computer networks. Approximately 150 students were enrolled in this course during the Spring 2020 semester, and had to quickly transition to an equivalent online experience. At NYU Tandon ECE, we relied heavily on GENI [3], a network- ing research testbed comprising resources located at universities and other sites around the United States, to offer a remote equiva- lent of the same lab content students would normally complete in person. In this white paper, we describe: • Resources that are available for educators interested in using GENI to teach computer networks, wireless communications, and network security. • Our experiences in the Spring 2020 semester, when we transi- tioned 150 students from a small-group in-person lab setting to an online lab setting, and outstanding challenges that remain to be addressed. 2 THE GENI TESTBED GENI [3] is an open infrastructure for networking and distributed systems research and education that spans multiple testbeds around the United States. In the educational resources described in this paper, we use two types of GENI testbeds. For experiments that do not require special wireless hardware, we use the InstaGENI [2, 4] testbeds. InstaGENI allows experi- menters to reserve, on demand, a topology of Linux virtual ma- chines in any network configuration, at one of more than several dozen sites located around the United States. An example of an Figure 1: Sample InstaGENI topology. This is the topology for a lab experiment on the spanning tree protocol, de- scribed in Table 1. InstaGENI topology for one of our lab experiments is illustrated in Figure 1. Experimenters configure and reserve their experimental resources using a simple browser-based interface, then log in to the hosts in their topology using SSH and execute experiments using standard Linux commands and utilities. For experiments that require WiFi or software defined radio hardware, we use GENI wireless partner testbeds: ORBIT [6] and WITest [1]. These sites include multiple smaller “sandbox” testbeds, which are reserved by experimenters in advance for use on a time- shared basis. Each of the smaller testbeds includes Linux hosts equipped with WiFi, software defined radio, or other wireless hard- ware devices. Various other platforms are available for teaching computer networks, including simulation platforms, in-house networking lab facilities, and solutions where the student’s PC (or a VM on a student’s PC) executes a network of virtual machines. Some benefits of the GENI platform specific to the online learning experience include: Software requirements: a terminal and a browser. Some students who use Windows on their personal machines may have to install a terminal application. Some lab exercises also suggest the use of Wireshark. Otherwise, the GENI-based lab exercises do not require students to install any extra software on their personal devices, and the actual execution of experiments happens on remote hosts. Stu- dents may use lightweight personal devices, such as a Chromebook, or even a tablet that is equipped with a keyboard. In a pandemic sit- uation, where students may not have access to university computer labs or to other shared computing resources, it was important to us that the lab infrastructure be equally usable to students who only