The Elements of Kubernetes - Foundational Concepts for Apps Running on Kubernetes
summary
Gives concrete best practices: use resource limits, readiness and liveness probes to enable horizontal pod autoscaling, prefer the crash-only pattern instead of complex retry code (one example replaces a 700-line retry loop), and use init containers to wait for services (SQL Server image takes ~6 minutes on a 7-node ACS). Demonstrates tools and failures: Helm for declarative deploys, Service Catalog for bindings, Fluentd sent 1M logs and crashed a 3-node cluster, and sidecar patterns with Envoy. Watch if you want actionable knobs and tradeoffs.
AI-written summary · about
chapters (19)
- 0:00Intro
- 0:11Why we're here
- 1:16Building an app for Kubernetes
- 2:53Lots to figure out
- 3:53North Star
- 5:59I've seen the good & the bad
- 6:26I'm here to propose ideas
- 7:15Observability is golden
- 8:16Kubernetes observing your app
- 9:33You observing your app
- 11:50Crash-only software
- 15:20But sometimes you need it
- 19:00Loose coupling is better than tight
- 19:35What that might look like
- 22:02Tight coupling isn't always wrong
- 24:15Record your configuration
- 26:17Ask for the least
- 27:32Build on the shoulders of giants
- 29:49We need your thoughts
related talks
29mKubernetes Workload Resiliency in Action: Beyond Basics
29mK8s Complexity Tamed: One Interface To Rule Them All With CNCF Incubat...
37mKubernetes Wonderland: Adventures in Platform Building
06mLightning Talk: Cloud Native Design Patterns
28mOverview Of Challenges And Solutions For Orchestrating Applications To Multiple DC And Edge Clusters
36mUnderstanding Kubernetes Through Real-World Phenomena and Analogies
20mThe path to self-healing: Re-architecting for massive scale on kubernetes
33mLevel Up Your Developer Experience 🧭