DPDK performance optimization for edge computing scenario
summary
Shows concrete OpenStack + DPDK tuning: CPU pinning, hugepages, PMD binding, NIC offloads and affinity changes; host and VM flavors set for pure isolation and large page memory. Uses OVS‑DPDK and VPP for user‑space datapath and AF_XDP sockets to mix CNF and kernel‑stack apps. Measured results: small‑packet throughput more than doubles vs kernel, and layer‑2 forwarding holds <30 µs for 512‑byte frames in their tests.
AI-written summary · about
chapters (8)
- 0:00DPDK performance optimization for edge computing scenario
- 1:25Edge computing application scenarios
- 5:56Key technical features of edge computing-real-time solutions
- 6:28Key technical characteristics of edge computing-high- performance network solutions
- 11:22Optimize OpenStack OVS-DPDK
- 13:47Test results conventional business applications of virtual machines
- 14:33Test results-conventional business applications of virtual machines
- 14:52Test result-NFV network element L2 forwarding
related talks
20mCan OVS-DPDK be further optimised?
36mRunning High-performance User-space Packet Processing Apps in Kubernetes
28mDPDK vs ODP: A comparison
14mPerf Characteristics of Userspace & Kernel-Space Datapaths in Container Networking
18mOpen vSwitch with AF_XDP as a userspace datapath
23mMigrating from DPDK to AF_XDP for High-Performance Networking in...
02h 53mPreview Sessions
36mBringing vDPA to Life in Kubernetes
