Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore a 16-minute conference presentation from NSDI '25 that introduces state-compute replication, a novel approach to parallelizing high-speed stateful packet processing across multiple CPU cores. Learn how researchers from Rutgers University, Politecnico di Milano, New York University, and Queen Mary University of London address the critical challenge of scaling packet processing throughput as Moore's law slows down and network interface card speeds continue to increase. Discover the limitations of traditional state sharding methods, which process packets updating the same state on a single core and become bottlenecked by skewed flow size distributions in realistic network traffic. Understand how state-compute replication overcomes these limitations by enabling multiple cores to update state without explicit synchronization through a packet history sequencer running on network interface cards or top-of-rack switches. Examine experimental results using realistic data center and wide-area Internet traces that demonstrate linear scaling of packet-processing throughput with the number of cores, regardless of flow size distributions, across various realistic packet-processing programs. Gain insights into this breakthrough approach that promises to maintain high-speed packet processing performance in software-based systems despite hardware scaling challenges.
