How Container Networking Affects Database Performance
Talkat KubeCon + CloudNativeCon North America 2019 · on Wednesday, 2019-11-20· at 15:20· Room 6F - San Diego Convention Center, Upper Level·
summary
Percona XtraDB Cluster (3‑node PXC) on Kubernetes is benchmarked with sysbench/TPCC on a 4‑node lab: 10Gb internal NICs, 1Gb external, local SSDs, iperf for bandwidth. Switching from 1Gb to 10Gb raises read/write throughput from ~2,700 r/s to ~7,300 r/s; synchronous replication multiplies latency effects. Tests compare CNI plugins (best DB throughput seen with kube‑router, Calico second), SR‑IOV+Multus gives highest raw bandwidth but worse DB performance, and flannel shows occasional multi‑second stalls. Quality‑of‑Service must be Guaranteed (requests==limits) to reduce variance.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 0:50About Percona
- 3:16Benchmark Layout
- 4:51Database Setup
- 12:32Parameters
- 19:15Results
- 23:49QA Session
- 30:50Summary
related talks
34mComprehensive Performance Benchmark on Various Well-Known CNI Plugins
32mBenchmark Your Cloud Native Database
32mBenchmark Your Cloud Native Database
05mIgnite - How Fast Was My Kubernetes?
31mA Practical, Research‑Backed Introduction To Cloud‑Native Kernel‑Bypass Networking
30mNetwork Configurations for High-Performance Streaming Connectivity in K8s
34mTroubleshooting Low Latency Application on CNF Deployment
22mEmpowering Edge Software Stack on Arm with High Performance Container Networking Tech