Enhancing DNS Reliability in Kubernetes With Node Local DN...
summary
Node-local DNS cache runs as a DaemonSet across hundreds of clusters with tens of thousands of nodes and reduces CoreDNS upstream traffic, cutting request forward duration by ~95% and request response time by ~98.5% in New Relic’s DNSPerf tests on AWS and Azure, with 0% query loss at 500 and 5,000 QPS and near-zero loss at peak; cache misses forward to CoreDNS, the system falls back on CoreDNS if caches are deleted, and node-local DNS cache yields dramatic latency savings.
AI-written summary · about
chapters (10)
- 0:00DNS reliability challenges
- 1:25Agenda and DNS issues
- 3:13Node local DNS cache
- 5:22Kubernetes DNS design
- 6:27Design with node local cache
- 7:33Testing strategy and results
- 13:27Impact on core DNS metrics
- 14:51Operational paranoia and failure testing
- 16:40Lessons learned and recommendations
- 18:08Q&A
related talks
39mObservability to Resolution: The Journey Through a Production K8s Incident
38mCoreDNS: Intro and Deep Dive
37mDig Smart: Building a Reliable Cloud Native DNS for Modern Networks
38mIntro + Deep Dive: CoreDNS
29mDNS: Old Solution for Modern Problems
43mKubeCon: A Story of Two Service Discoveries: Bringing Legacy to Kubernetes
10mAvoiding DNS Pain
03mLightning Talk - Introducing CoreDNS Kubernetes Integration