Explore a 14-minute conference talk from OSDI '23 that introduces ShRing, an innovative approach to networking with shared receive rings. Delve into the challenges of multicore systems parallelizing incoming Ethernet traffic and the potential memory bandwidth issues that arise when scaling to hundreds of gigabits per second. Learn how ShRing addresses these problems by sharing receive rings among multiple cores, resulting in improved throughput and significantly reduced latency for NFV workloads. Discover the trade-offs between software synchronization costs and reduced memory footprint, and understand how ShRing can prevent overload conditions in high-speed networking scenarios.
Overview
Syllabus
OSDI '23 - ShRing: Networking with Shared Receive Rings
Taught by
USENIX