Kubernetes, Decisions
summary
Seven concrete decisions for running Kubernetes at scale: choose tenancy model (trusted, trusted soft isolation, or untrusted), decide cluster count (customers hit 500 microservices / 2000–5000 containers in one cluster and pay the blast radius), prefer fewer clusters when constraints allow, push tenancy into application design or isolate untrusted workloads (example: move from K-native to Lambda for UDFs), use policy engines (Kyverno/OPA), consider purpose-built OS (Bottlerocket/Flatcar), and start multi-cluster connectivity with cloud load balancers before attempting mesh federation.
AI-written summary · about
chapters (6)
- 0:00Introduction and speaker background
- 3:37Multi-tenancy challenges in Kubernetes
- 19:16Cluster count and isolation boundaries
- 25:02Purpose-built container operating systems
- 29:36Developer productivity vs platform control
- 32:46Multi-cluster connectivity strategies
related talks
20mDesigning a Multitenancy Kubernetes Strategy That Fits Your Org
32mLaying the Foundation: Real World Kubernetes Deployment Patterns
54mKubernetes for the konfused, a strategy toolkit for normals
35mKubernetes, Lessons Learnt
37mMulti-Tenancy in Kubernetes: Best Practices Today, and Future Directions
39mKubernetes is awesome – but …
33mKubernetes Multi-tenancy: Principles and Practices for Large Scale Internal Deve...
05mLightning Talk: What Makes a Good Multi-tenant Kubernetes Solution?