Scott Shenker
Location: (Berkeley, CA)
Personal Research Web Page: http://www.eecs.berkeley.edu/Faculty/Homepages/shenker.html
Keywords: Internet Architecture, Network Design, Datacenter Infrastructure
Posted on: Wednesday, May 19th, 2010
Broad Research Area: Networks / Operating Systems
Research Interests:
Over the years my research interests have covered a broad set of topics, ranging from system design to theoretical formulations to economic analysis. My current interests fall into five main areas:
* New Internet architectures: in collaboration with Hari Balakrishnan, Nick Feamster, Brighten Godfrey, Teemu Koponen, Nick McKeown, Guru Parulkar, and Jen Rexford, my research group is pursuing a new approach to Internet architecture that is built around a minimal framework that enables innovation in the rest of the architecture. I am actively looking for fellows eager to work in this area.
* Software-defined networking: networking is on the verge of a revolution, going from complicated switches/routers running specialized control algorithms (routing algorithms, etc.) to a new model where simple switches/routers follow instructions sent from management applications built on a general control platform (such as NOX, a network operating system). My research group is exploring this new approach in collaboration with colleagues at Stanford (Nick McKeown) and Nicira Networks (Martin Casado and Teemu Koponen). We are investigating both the control platform itself and new management applications (such as traffic engineering for the WAN, control mechanisms for datacenters, etc.).
* New network algorithms and protocols: As always, I am interested in new network mechanisms of almost any variety, from network monitoring to congestion control to intradomain routing to interdomain policy enforcement.
* Systems infrastructure for datacenters: Datacenters are the new unit of computing, and my research group (in collaboration with many colleagues here at Berkeley) are exploring the basic pieces of infrastructure needed to fully exploit the vast computation power residing in today’s large datacenters. So far we have focused on new resource allocation mechanisms (Nexus) and a new programming framework (Spark).
* Hypervisor-based security tools: We are exploring a range of security measures that can be implemented in today’s hypervisors. Two examples of this are Practical Taint Tracking (which has improved the performance of taint tracking by almost an order of magnitude) and Virtual Dedicated Console (which enables secure access to online services).
Contact Information:
shenker@icsi.berkeley.edu
