Talk by Tim Brecht, University of Waterloo: The Approaching Tsunami of HTTP Streaming Video

Speaker: Tim Brecht, University of Waterloo

Host: Andy Warfield

Title: The Approaching Tsunami of HTTP Streaming Video

Abstract:

Rapidly growing amounts of Internet streaming video traffic are being delivered to viewers by companies like Apple, Adobe, Akamai, Netflix and Microsoft using HTTP. HTTP is being used because it is simple, it works through firewalls, and the existing ubiquitous infrastructure that consists of web servers, caches, Content Distribution Networks (CDNs) and clients can be easily leveraged. While this infrastructure (the HTTP ecosystem) does work, it has been designed, tuned and optimized to serve primarily small files that exhibit a high degree of locality, rather than large video files with a long tail of content that is viewed only a few times. In this talk I describe some of the work we are doing to understand and improve the performance of the HTTP ecosystem while serving streaming HTTP video. I will first describe a methodology for generating HTTP streaming video workloads and how we have used one of these workloads to benchmark existing web servers. I will then show how a simple modification to one of these servers has resulted in a twofold increase in throughput and explain why we think more work needs to be done. This is work being done in conjunction with Jim Summers, Bernard Wong, Tyler Szepesi and Ben Cassell all at Waterloo and Derek Eager at the University of Saskatchewan.

Bio:

Tim Brecht is an Associate Professor in the Cheriton School of Computer Science at the University of Waterloo. He has previously held positions as an Associate Professor at York University (in Toronto), a Visiting Scientist at IBM's Center for Advanced Studies, a Research Scientist with Hewlett Packard Labs, and Visiting Professor at Ecole Polytechnique Federal de Lausanne (EPFL). Research interests include: empirical performance evaluation, Internet systems services and applications, parallel and distributed computing, operating systems, networking, and smart grid infrastructure. His research has been recently recognized with an NSERC Discovery Accelerator Grant.