Research Projects:

  • Internet Sustainability: We explore data-driven models and architectures towards a more sustainable Internet in light of the ever-increasing traffic demand, infrastructure deployment and maintenance cost, and competitiveness in the Internet ecosystem. Our initial results include the development of algorithms for incremental and sustainable upgrades of ISP backbones and innovative use of in-network storage. [2016-present, collaboration with Yale University, Trinity College Dublin, University of Freiburg, MIT, and Akamai].

  • Internet and Content Delivery Analytics: We develop novel and scalable techniques to assess the state and health of the Internet and to improve content delivery in a rapidly changing Internet. We push the envelope in Internet measurement by relying on a voluminous amount of data collected from both public and private vantage points and introduce new techniques to map peering interconnections to the level of a single building. We also exploit the distributed platform of a large content delivery network, composed of thousands of servers around the globe, to assess the performance characteristics of the Internet's core. We investigate the cost-performance tradeoffs the different interconnection service offerings that are available to networks in today's Internet in one and the same colocation facility would go a long way towards putting this debate on scientifically solid foundations. In the process, we designed a fully decentralized, open-source analytics system for network traffic data that relies on smart partitioning storage schemes to support fast join algorithms and efficient execution of filtering queries. [2014-present, the projected is hosted at MIT and TU Berlin, and it is a collaboration with Akamai].

  • Enabling ISP and CDN Collaboration: We develop protocols and systems to enable the collaboration between ISPs and CDNs. We design and evaluate a service provided by an ISP, called PaDIS, to improve end-users experience based on the network information and end-user location available to an ISP, as well as server and path diversity information gathered by an ISP. We also introduce Content-aware Traffic Engineering (CaTE) which dynamically adapts server selection for CDNized content by ut ilizing server and path diversity and without changing routing. Moreover, we show how to enable CDN and ISP collaboration, with our NetPaaS prototype, in light of recent CDN-ISP alliances and advances in network function virtualization. Our results show that CDN-ISP Collaboration leads to a win-win situation for both parties and improves end-user experience. [2008-today]

  • Content Cartography and DNS: We propose a lightweight and fully automated approach to discover hosting infrastructures based only on DNS measurements and BGP routing table snapshots. Our classification enables us to derive content-centric AS rankings that complement existing AS rankings and sheds light on recent observations about shifts in interdomain traffic and the AS topology. In a parallel effort we undertake a large scale study to assess DNS performance in the wild and we highlight the implications of DNS deployment to end-users, ISPs and applications. [2009-2017]

  • IXP Data Analysis: Large Internet Exchange Points (IXPs) are responsible for exchanging more than 25 Petabytes of data daily but have received very little attention from the research community. We have established a pioneer research project to assess the importance of IXPs for today's Internet ecosystem. Our measurement results show that large IXPs are unique vantage points to observe trends in peering and routing, content delivery, Internet services and applications, and revisit AS-level view of the Internet. [2012-today]

  • Delay Tolerant Bulk Transfers on the Internet: We design and evaluate scheduling algorithms, some assisted by network-attached storage, to transfer delay tolerant bulk data over the Internet with the most cost effective way. [2008-2013]

  • Selfish Neighbor Selection: We re-examine the problem of overlay network creation, taking into consideration the existence of selfish overlay nodes. We develop a general game-theoretic framework that provides a unified approach to modeling Neighbor Selection procedures on behalf of selfish nodes. To capitalize on the substantial performance improvement of best response wirings for overlay nodes, we design, deploy and evaluate, EGOIST, a Selfish Neighbor Selection inspired prototype. We also show the benefits Selfish Neighbor Selection may offers to applications. [2006-2013]

  • Distributed Facility Location: We design and evaluate distributed algorithms for scalable and efficient service deployment and migration. [2006-2013]