Where the time goes
Our research interests span several areas of experimental systems with a focus on distributed and parallel computing, operating systems, and to some extent, networking. We focus primarily in large-scale and globally-distributed services, implemented at the application layer and following a cooperative, self-organizing model where hosts contribute resources to support a commonly needed service.
To be successfully deployed in a complex and open setting such as the Internet, large-scale distributed systems must be mostly self-manageable and they must tolerate dynamic membership, high heterogeneity, and large dynamism in resource demand and availability. Our work addresses these challenges using an experimental approach in the context of a variety of cooperative application classes - from content sharing to overlay multicast and distributed systems based on vehicular ad-hoc networks (VANETS).
More recently, our work is increasingly focused on ensuring sustainable scalability for large-scale distributed computing. I believe that any solution to systems problems in this context must be carefully crafted so as not to impose unbearable demands on the environment or the systems administrators at risk of curtailing further growth. For example, we have started to explore a new approach to the design and implementation of distributed systems -- one that focuses on minimizing aggregated control and administrative overhead by strategically reusing environment~s views and/or recycling previously gathered measurements. We call this approach 3R for Reduce, Reuse, Recycle, highlighting the analogy between our natural habitat and the artificial, Internet environment shared by these systems.
We sponsor a number of undergraduate student research projects in the area of distributed systems. Please contact us if you are interested in how to get involved with our research.
Some of our ongoing projects include ...
- Sustainable Scalability for Globally-Distributed Systems
- Reusing CDN Monitoring Information
- Scalable performance detouring through information reuse
- Reusing P2P for early detection on of network problems
- Reducing P2P cross-ISP traffic while improving users' performance.
- Cooperative Data-Sharing Services
- Handling churn in less-structured P2P systems
- Looking at the server side of peers
- Structured or Unstructured Overlays?
- Streaming Overlay Multicast
- Nemo - Structurally resilient overlay multicast
- FatNemo - FatTrees for bandwidth intesive applications
- Magellan - Achieving fairness without performance penalties
- C3 - Car-to-Car Cooperation for vehicular ad-hoc networks
- STRAW - Realistic vehicular mobility model
- VFN - Virtual Ferry Networking for exploiting emergent behavior in vehicular networks
- SWANS++ - Unified extensions and upgrades for the SWANS wireless network simulation
- Center of Excellence for Experimental Systems in Education