Fast and Smart Network Resource Management for Datacenters and Beyond

Time: Friday, November 17, 2017 - 2:00pm - 3:00pm
Type: Seminar Series
Presenter: Mohammad Alizadeh; Department of Electrical Engineering and Computer Science - Massachusetts Institute of Technology
Room/Office: Becton 035
Becton Seminar Room
15 Prospect Street
New Haven, CT 06511
United States

Department of Electrical Engineering Seminar

Mohammad Alizadeh
TIBCO Career Development Assistant Professor
Computer Science & Artificial Intelligence Laboratory
Department of Electrical Engineering and Computer Science
Massachusetts Institute of Technology (MIT)

"Fast and Smart Network Resource Management for Datacenters and Beyond"

Abstract: Modern networked systems give rise to many challenging resource management problems, from congestion control in datacenters to bitrate adaptation for video streams to scheduling workloads on computer clusters. In this talk, I will describe some of our work on network resource management systems and algorithms across two paradigms: (1) a classical paradigm, characterized by simplified system models, low-level design goals, and fixed algorithms that achieve these goals in specified conditions; (2) a learning paradigm, characterized by unknown and uncertain system dynamics, high-level resource management goals, and algorithms that learn and adapt to their environment and workload.

For the classical paradigm, I will describe some of our work on packet transport mechanisms for datacenters that enable the construction of large-scale fabrics which provide 100s of Tb/s of bandwidth, microsecond-level packet latencies, and resilience to link failures. For the learning paradigm, I will discuss our recent work on systems that learn to make resource management decisions entirely from experience using modern deep reinforcement techniques. I will demonstrate the power of this approach for problems such as adaptive bitrate selection for video streaming and scheduling jobs in computer clusters.

Bio: Mohammad Alizadeh is the TIBCO Career Development Assistant Professor in the Department of Electrical Engineering and Computer Science (EECS) at MIT, and a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL). He completed his graduate studies at Stanford University, earning his Ph.D. in electrical engineering in 2013, and then spent two years at Insieme Networks, a datacenter networking startup, and Cisco Systems before joining MIT.

Mohammad's research interests are in the areas of computer networks and systems. His current research centers on network protocols and algorithms for large-scale datacenters, programmable switching architectures, and learning-based networked systems. He is also broadly interested in performance modeling and analysis of computer systems and bridging theory and practice in computer system design. Mohammad's research has garnered significant industry interest. His work on datacenter transport protocols has been implemented in the Linux and Windows operating systems and has been adopted by leading network operators such as Microsoft and Google; his work on adaptive network load balancing algorithms has been implemented in Cisco's flagship datacenter switching products. Mohammad is a recipient of several awards, including the SIGCOMM Rising Star Award, Alfred P. Sloan Research Fellowship, Google Faculty Research Award, Facebook Faculty Award, SIGCOMM Best Paper Award, and Numerical Technologies Inc. Prize and Fellowship.

Hosted by: Professor Leandros Tassiulas

Friday, November 17, 2017
Becton Seminar Room
2 PM