Dr. Azer Bestavros
Boston University

Abstract

In many emerging settings where users are empowered to make autonomous resource acquisition decisions, and in which infrastructure providers and users are interested in selfishly maximizing their own utilities, resource management must be viewed through a game-theoretic (as opposed to a global optimization) perspective. In this talk, I will present such perspectives in three distinct settings: overlay network creation, cloud resource colocation, and shared bandwidth arbitration.

For overlay networks, I will present "Selfish Neighbor Selection" (SNS) as a game-theoretic connectivity management framework for folding new arrivals into an existing overlay, and for re-wiring to cope with changing network conditions. I will show that under typical resource constraints, SNS yields Nash-like equilibria that also provide high global performance. I will present experimental results showing the properties of stable SNS wirings on synthetic and real Internet topologies, as well as results we obtained by deploying Egoist - a PlanetLab SNS prototype system.

For cloud computing, I will present "Colocation Games" (CG) as an economically-sound framework upon which emerging cloud architectures could be implemented. CGs enable the modeling and analysis of the dynamics that result when rational, selfish parties interact in an attempt to minimize the individual costs they incur to secure the shared cloud resources necessary to support their application QoS or SLA requirements. In addition to various game-theoretic results, I will overview implementation considerations as well as results from experimental evaluations.

For shared bandwidth arbitration, I will present "Trade and Cap" (TC) as an economics-inspired mechanism that incentivizes users to voluntarily coordinate their consumption of a shared resource so as to converge to what they perceive to be an equitable allocation, while ensuring efficient resource utilization. Under TC, rather than acting as an arbiter, service providers act as enforcers of what the community of rational users decides is a fair allocation. In addition to presenting the analytical underpinnings of TC and results from trace-driven simulations, I will briefly discuss implementation considerations for last-mile bandwidth arbitration.

This work was pursued primarily in collaboration with Jorge Londono, George Smaragdakis, and Nikos Laoutaris.

Bio

Azer Bestavros (PhD'92, Harvard U) is Professor in the Computer Science Department at Boston University, which he joined in 1991, and which he chaired from 2000 to 2007. Azer's research interests are in the broad areas of networking and real-time embedded systems. His contributions include his pioneering the distribution model adopted years later by CDNs, his seminal work on Internet and web characterization, and his work on compositional certification of networked systems and software. Funded by grants totaling over $16M from various government agencies and industrial labs, his research work yielded 13 PhD theses, over 80 masters projects, 4 issued patents, 2 startup companies, over 160 refereed papers, and over 3,700 citations. He is the chair of the IEEE Computer Society TC on the Internet, and received distinguished ACM and IEEE service awards.