Scaling Open vSwitch with a Computational Cache

Explore a conference talk that presents innovative techniques for scaling Open vSwitch (OVS) to support hundreds of thousands of OpenFlow rules. Learn about the integration of NuevoMatch, a neural network-based packet classification algorithm, with OVS to significantly improve its performance and scalability. Discover how the researchers overcame the challenge of slow rule updates in NuevoMatch, achieving a speed-up of over three orders of magnitude. Examine two design options for integrating NuevoMatch with OVS: as an additional caching layer and as a complete replacement for OVS's data-path. Analyze the comprehensive evaluation results, which demonstrate substantial speedups and improved rule update capabilities. Gain insights into the potential impact of this research on virtual switch implementations and network performance optimization.