A Practical, Research‑Backed Introduction To Cloud‑Native Kernel‑Bypass Networking
summary
Baseline Linux TCP networking tops at 0.6–0.7 MPPS with iperf3 on 10Gb links while a user‑space path using DPDK and testpmd hits 10Gb line rate at 64B frames; a user‑space TCP stack (lwIP or the new IAP) with memcached shows large gains and IAP reaches ~70M RPS on 32 cores versus 4.8M RPS for Linux. Vhost user reaches ~20–50M PPS, SR‑IOV attains line rate, and Kubernetes integration options are secondary NICs, extending CNIs, or new CNI plugins.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 1:07High-speed networks underutilize potential
- 5:00Kernel bypass basics and DPDK
- 8:57User-space TCP/IP stacks
- 11:02Benchmarking kernel bypass TCP/IP
- 15:14Summary of kernel bypass benefits
- 15:35Container networking challenges
- 18:02Kernel bypass container setup
- 21:34SR-IOV hardware virtualization
- 22:59Summary of container kernel bypass
- 23:20Cloud-native kernel bypass in Kubernetes
- 27:14Feasibility and future of cloud-native bypass
- 28:08Q&A
related talks
20mNetKernel: Making Network Stack Part of the Virtualized Infrastructure
34mCan the Linux networking stack be used with very high speed applications?
17mRearchitecting the TCP Stack for I/O-Offloaded Content Delivery
17mOpening Up Kernel-Bypass TCP Stacks
26mTuning Your Cloud: Improve Global Network Performance for Application
14mHigh-throughput and Flexible Host Networking for Accelerated Computing
37mSPDK Based User Space NVMe Over TCP Transport Solution
36mRunning High-performance User-space Packet Processing Apps in Kubernetes
