Krill Prize 2015
Improving the Internet Infrastructure
Despite the crucial societal role of the Internet, the core Internet infrastructure, i.e., the protocols that handle the fundamental networking tasks such as routing, congestion control, etc. (TCP/IP, DNS,BGP), has remained essentially stagnant for over two decades. Consequently, today’s Internet often fails to cope gracefully with modern-day challenges, e.g., alarming security vulnerabilities, exponential growth in scale and complexity, increasingly heavy traffic loads, frequent equipment failures, and more. My research aims to (1) understand the existing Internet infrastructure and the fundamental tradeoffs that should guide the design of the future Internet, and (2) “fix” today’s infrastructure and design lasting, long-term solutions. To accomplish this, my research
• takes a principled approach, drawing inspiration from theory (e.g., algorithmics, distributed computing, learning, microeconomics) in search of practical approaches that go beyond context-specific, ad hoc solutions.
• breaks traditional disciplinary barriers (both within and without Computer Science) to
tackle the above challenges from a holistic point of view.
I next briefly discuss several of my research agendas along the above lines.
New Paradigms for Securing the Internet. The Internet’s core protocols were not designed with security in mind and are, consequently, alarmingly vulnerable to attacks. While secure versions of these protocols were proposed over a decade ago, adoption is either not even on the horizon or frustratingly slow. A particularly stubborn problem is that of securing routing between the smaller networks that make up the Internet (e.g., Google, AT&T, Hebrew University, etc.). My research focuses on eliminating operational, political and economic obstacles to securing the Internet infrastructure and, in particular, on exploring new paradigms for securing Internet routing. New Architectures for High-Performance Networks. The rapid growth of Internet services (from e-mail to social networks and streaming media) is placing tremendous demands on communication networks. To meet this challenge, my research aims to devise high-performance network architectures and, in particular, new network topologies and new protocols for routing, congestion control, traffic engineering, and more. To accomplish this, I strive to turn deep ideas from theory (graph theory, algorithmics and computational learning theory) into an operational reality. Beyond the important practical implications of this research direction, it also motivates new theoretical questions. New Horizons. Software-Defined Networking (SDN) is a new paradigm that is revolutionizing networking, fueled by adoption by major technology companies, e.g., Google and Amazon. SDN enables logically-centralized control over network devices through software that operates independently from the network hardware. My research explores exciting SDN-motivated questions regarding the “right” abstractions for “programming networks” and how to defend the network against harmful SDN programs. Answering these questions involves exploring the interface of traditional networking research with other disciplines in Computer Science, e.g., programming languages, software analysis and databases.