About Me

I am a professor in the Electrical and Computer Engineering Department at Carnegie Mellon University and at Carnegie Mellon University-Africa. My group focuses on developing models and designing algorithms and frameworks to improve the performance and availability of computer networks. In particular, my group explores the critical role that network state plays in determining network performance and availability with the state’s inherent semantics and emergent state management techniques by investigating designs and algorithms to more holistically understand and manage this state. More recently, we have applied this approach to addressing the digital divide, improving microservices/clouds, managing software defined networks, and rethinking CDN designs. Our designs and systems have been deployed at web scale companies, adopted by opensource systems, and acquired by a large hardware manufacturer.

Hiring: I am actively recruiting MS, PhD, and Post-Docs.

1. Post-Docs: I am looking to hire 2 post-doc researchers who specialize in programmable networks (e.g., eBPF, P4) or network measurements, starting in mid-2024. We are actively seeking highly motivated PhD students interested. Send me an email.
2. For CMU-MS students, you can find project descriptions here and on ECE's SPT portal.

News

Current Research Group

• Lucas Castanheira, PhD
• Isabel Suizo, PhD (Co-advised with Justine Sherry)
• Mugdha Deokar, Research Associate
• Yu-Ting Wang, PhD (Visiting Student from National Taiwan University of Science and Technology)
• Jingyang Men, MS

Research

Broadly, my group focuses on improving network performance and availability. In data centers and clouds settings, we work on designing abstractions and frameworks for managing programmable substrates and eliminating existing complexity. At the edge, we work on developing measurement techniques and models to understand, debug, and eliminate the complex interactions between protocols that result in performance and availability issues.

• Democratizing Web Performance
  • African Internet Observatory: Collection of assessment probes, statistical measurement methods, and simulations to analyze the internet ecosystem in Africa. More can be found here.
  • Data-driven Protocol Optimizations: [Hydra, EuroSYS MLSYS'23] [TCP-QOS, IWQoS'23] [ConfigTron, NSDI'22] [ZeroDownTimeUpdates, SIGCOMM'20] [ConfigTron, HotCloud'17] [YVideo, IMC'16]
  • Web Performance in Developing Regions: [JSCapsules, SIGMETRICS'23] [WebMedic, HotMobile'21] [AnazlyingMobileBottlenecks, CCR'20] [MobilePerfInPakistan, IMC'16]
  • Quantifying and Analyzing Configuration Heterogenity: [WebPerfOpt, HotMobile'21] [InspectorGadget-TCP, TMA'20] [InspectorGadget-HTTP, MobileQoE'18]
  • Dissecting QUIC Performance: [ProductionQUIC, WWW'21]
• Abstractions for Programmable Infrastructures
  • Fault Tolerant SDNs: [Tardis, SoSR'21] [Mozart, SoCC'19] [Chopin, APNet'17] [LegoSDN, SoSR'16]
  • Debugging and Diagnosis: [uView, CoNEXT'23], [FoxHound, EuroSYS'23], [P4-InTel, CoNEXT ECNP'19] [P4Visor, CoNEXT'18] [Dapper, SoSR'17] [Armageddon, HotNets'15] [YTrace, arXiv] [Yahoo Data Centers, arXiv]
  • Storage Abstractions: [Nimble, INFOCOM'23] [SCC, EuroSYS'19] [Hermes, CoNEXT'17] [ Mecury, SoSR'17] [ SwitchVisor, CAN'17]
• Self Managing Networks: Machine Learning and Networks
  • Self Configuring Data Centers: [DeepConfig, ArXiv] [DeepConfig, NetAI'18] [DeepConfig, NeurIPS MLSYS'18]
  • IoT Security: [ IoT Vision, IoTSP'17] [ Learning MUD, SoSR'19] [ Verifying MUD, TDSC'20] [Pot2DPI, CoNEXT CAN'17]

Software and Datasets

• Data Center Characteristics [IMC'2010]: data set
• LegoSDN (Fault Tolerant Controller) [SoSR'16]: source code
• Dapper (Data Plane diagnosis of TCP) [SoSR'17]: source code
• P4Visor [CoNEXT'18]: source code
• Hermes [CoNext'17]: Coming soon!

Funding: Grants & Gifts

• Google Faculty Research Awards, 2018 - 2019.
• NeTS: Small: Data Driven Mobile Web Performance., 2018 - 2022.
• NeTS: Small: Network-centric IoT Security., 2018 - 2022.
• NSF CAREER: Bug Tolerant Networking: Enabling Highly-Available Network Infrastructures through Semantic Transformations., 2018 - 2023.
• Democratizing Web Performance: Principled Measurements and Optimizations for Performance in Developing Regions, 2018 - 2019.
• NSF Medium:Towards Finer-Grained Cloud Computing., 2014 - 2019.
• Facebook Faculty Award, 2015, 2016.
• Google Cloud Resource Usage Grant, 2015.
• Yahoo! FREP, 2014, 2015.
• Yahoo! ACE, 2014.
• 2014 Equipment Gift from Yahoo, 300 servers.
• Facebook Gift, 2014-2016.
• Duke Bass Connections, 2014.

Selected Publications

Full list of publications can be found here.

1. Configanator: A Data-driven Approach to Improving CDN Performance.
   Usama Naseer, Theophilus A. Benson
   [Proceedings of USENIX Networked Systems Designs & Implementation (NSDI), Renton, Washington, Apr 2022]
   
2. Dissecting Performance of Production QUIC.
   Alexander Yu, Theophilus A. Benson
   [Proceedings of The Web Conference (WWW), Ljubljana, Slovenia, Apr 2021]
   
3. Zero Downtime Release: Disruption-free Load Balancing of a Multi-Billion User Website.
   Usama Naseer, Luca Niccolini, Udip Pant, Alan Frindell, Ranjeeth Dasineni,Theophilus A. Benson
   [Proceedings of ACM Special Interest Group on Data Communication (SIGCOMM), New York, New York, Aug 2020]
   
4. In-Network Compute: Considered Armed and Dangerous.
   Theophilus A. Benson
   [Proceedings of ACM Workshop on Hot Topics in Operating Systems (HotOS), Bertinoro, Italy, May 2019]
   
5. Efficient and Safe Network Updates with Suffix Causal Consistency.
   Sheng Liu, Theophilus A. Benson, and Michael Reiter
   [Proceedings of ACM European Conference on Computer Systems (EUROSYS), Dresden, Germany, Mar 2019]
   
6. Performance Characterization of a Commercial Video Streaming Service.
   Mojgan Ghasemi, Partha Kanuparthy, Ahmed Mansy, Theophilus Benson and Jennifer Rexford
   [Proceedings of ACM Internet Measurement Conference (IMC), Santa Monica, CA, Nov 2016]