Spreading Apps, Controlling Traffic & Optimising Costs in Kubernetes
summary
Explains how to spread 10 replicas across 3 AZs to get a 3‑4‑3 distribution using pod topology spread constraints, contrasts that with anti‑affinity pitfalls, and shows traffic steering via Istio locality weighted load balancing (examples: 60/40 and 80/10/10 per AZ, and forcing 98% to one AZ to validate rules) to cut cross‑zone egress costs. Describes node scaling with Karpenter consolidation, which respects topology constraints, provisions nodes via EC2 API, and currently targets AWS only.
AI-written summary · about
chapters (9)
- 0:00Start
- 1:43Long, long time ago...on a Black Friday
- 4:00High Availability
- 5:36Scaling Nodes
- 8:04Pod Anti-Affinity Rules
- 9:43Pod Topology Spread
- 11:50Istio Service Mesh
- 13:09Istio Destination Rules
- 16:53Topology Aware Hints
related talks
22mSpreading Apps, Controlling Traffic & Managing Costs in Kubernetes
45mKubernetes Resiliency
59mKubernetes - going beyond the basics
27mImproving Network Efficiency with Topology Aware Routing
23mOptimizing Storage Assignment via Pod Scheduling Under Disturbance Factors
24mK8s Pod Scheduling - Deep Dive
32mLaying the Foundation: Real World Kubernetes Deployment Patterns
46mThe Journey of Migrating 2500 Apps to Multi-Cluster Kubernetes and Zero Trust Istio