Andreas Wundsam

Andreas Wundsam

International Computer Science Institute
Networking Research Group
1947 Center Street
Berkeley, California 94704

Phone: +1 (510) 666 - 2966

E-Mail: andi _AT_ icsi _DOT_ berkeley D_OT edu

Research Interests:

I am currently a postdoc in the Networking Research Group at the International Computer Science Institute (ICSI) and visiting postdoc at NetSysLab of UC Berkeley. I received my Ph.D. from TU Berlin, advised by Anja Feldmann. My research interests include:

Improving Control and Troublehooting of Operational Networks

As networks and the systems that run on them get more and more complex, our abilities to monitor and debug them remain painfully inadequate. Many difficult-to-find bugs stem from complex interactions in large scale networks, combined with unpredicted behavior from users over an extended period of time, and can thus not easily be found in simulators or testbeds. We leverage Network Virtualization to propose several new approaches to overcome these limitations.

OFRewind

Mirror VNets

Network Virtualization

The Internet arguably is at its core currently at an impasse. Except for ever-increasing bandwidth, technical innovations get deployed only very sparsely any more, even if their usefulness is largely undisputed. Failed or stuck examples include IPv6, DiffServ, DNSSec, Inter-Provider QOS and many others. Additionally, modern applications pose ever requirements on the net that are ever more demanding and sometimes outright contradictory.

Network Virtualization is a key enabler in making practical deployments of such approaches feasible. It enables different entire networks to coexist on a shared physical infrastructure, ideally while preserving full isolation and enabling administrative access by independent actors. With my colleagues, I am working on a possible architecture for Network Virtualization and exploring applications within the scope of debugging.

Testbeds

Flow routing

The data rates provisioned by broadband Internet access connections continue to fall short of the requirements posed by emerging applications. Yet the potential of statistical multiplexing of the last mile broadband connections remains unexploited even as the average utilization of these connections remains low. We propose flow-based routing in community networks as a solution to the quest of improving the end-user QoE in times of peak bandwidth demand.

Other work

I am not exclusively a research guy :). During my studies and after, I have worked for a number of startups and other small software companies, mostly dealing in Java based Software Engineering. A selection of companies I have been involved in:

Personal interests

Last update: Sept 06, 2011