Abstract

Network simulation has long been used as a tool for evaluating network algorithms such as congestion control. However, configuring a simulator to faithfully mimic the target network is challenging. We articulate the goal of ”learning the network” to recreate real-world network scenarios, i.e., mimicking the end-to-end behaviour of network paths based on telemetry in the form of input-output packet traces. We then present our proposal, iBox, or Internet in a Box, to learn a model of the network path, helping recreate the end-to-end behaviour of the target network when presented with a new input. We discuss both networking domain knowledge-based and machine learning-based approaches to iBox, pointing out the trade-offs between these, the challenges presented by each, and the opportunity to meld the two. We present results from an Internet research testbed and from controlled experiments that show the promise of iBox.

Bio:

Venkat Padmanabhan is Deputy Managing Director at Microsoft Research India in Bengaluru. His research interests are broadly in networked and mobile systems, and his work over the years has led to highly-cited papers and paper awards, technology transfers within Microsoft, and also industry impact. He received the Shanti Swarup Bhatnagar Prize in 2016, and several test of time paper awards from ACM SIGMOBILE , ACM SenSys, and ACM SIGMM . He was also among those recognized with the SIGCOMM Networking Systems Award 2020, for contributions to the ns family of network simulators. Venkat holds a B.Tech. from IIT Delhi (from where he received the Distinguished Alumnus award in 2018) and an M.S. and a Ph.D. from UC Berkeley, all in Computer Science, and has been elected a Fellow of the INAE , the IEEE , and the ACM . He is an adjunct professor at the Indian Institute of Science and was previously an affiliate faculty member at the University of Washington. He can be reached online at http://research.microsoft.com/~padmanab/.