Skip navigation
Previous searches
    Suggested searches

      PacketMill: Toward Per-Core 100-Gbps Networking

      We present PacketMill, a system for optimizing software packet processing, which (i) introduces a new model to efficiently manage packet metadata and (ii) employs code-optimization techniques to better utilize commodity hardware.

      Our evaluation results show that PacketMill increases throughput (up to 36.4Gbps or 70%), reduces latency (up to 101 μs or 28%), and enables nontrivial packet processing (for example router) at about 100Gbps, when new packets arrive more than 10 times faster than main memory access times, all while using only one processing core.  

      Full abstract in ACM Digital library

      Authors

      Alireza Farshin, Tom Barbette, KTH Royal Institute of Technology

      Amir Roozbeh, Ericsson Research

      Gerald Q. Maguire Jr, Dejan Kostić, KTH Royal Institute of Technology

      Presented at ASPLOS 2021: 26th ACM International Conference on Architectural Support for Programming Languages and Operating Systems, April 2021

      Download pdf

      PacketMill: Toward per-core 100Gbps networking

       pdf