Tune It Up! Enabling Low Latency in Kubernetes Clusters
Talkat KubeCon + CloudNativeCon North America 2020 Virtual · on Thursday, 2020-11-19· at 17:40· Intrado Virtual Event Platform,·
summary
Demonstrates turning Kubernetes nodes into low‑latency hosts by installing a real‑time kernel and applying node tunings via machine-config-operator, managing tuned profiles with node-tuning-operator and tuned, and orchestrating kubelet settings (CPU manager, topology manager), SR‑IOV device plugins and hugepages with the performance operator. Live demos show machine configs changing Kubelet files, tuned applying isolated‑CPU kernel args, and pods receiving NUMA‑aligned CPUs plus SR‑IOV VFs, producing deterministic allocations and measurable latency improvements.
AI-written summary · about
chapters (12)
- 0:00Intro
- 1:18Feature use case
- 3:26Operators
- 3:59Real Time Kernel
- 8:11Micro Demo
- 11:18Node Tuning Operator
- 14:10CPU manager
- 14:59Device manager
- 15:36Topology manager
- 16:28Topology-aware resource allocation
- 17:16Example: Not-NUMA-Aligned resources
- 21:48Performance Addon Operator as orchestrator
related talks
36mOptimizing Kubernetes for Low-Latency and Real-Time Containers
02h 26mKubernetes Performance Tuning Workshop
32mOptimizing Application Performance on Kubernetes
19mOpenShift for low latency and real time workloads
30mTuning and automating for Telco 5G containerized workloads
21mDo Not Disturb Mode: Fine-Tuning Resources for Latency Sensitive Workloads
31mRunning Low-Latency Workloads on Kubernetes
15mOptimizing Application Performance on Kubernetes