Why Kubernetes is Not Enough
summary
Farfetch runs an active-active global stack across three data centers and thousands of servers with petabyte-scale storage, and is migrating ~500 microservices into Kubernetes while doing ~25,000 builds/month and ~1,200 weekly deploys. Kubernetes provides autoscaling and rollout primitives but fails without solid CI/CD, contract testing, unified observability, security and cost visibility. The team builds blueprints as the single source of truth to drive automated provisioning, dependency wiring, Kafka topics, and RBAC, and emphasizes copyable staging clusters and 25,000 builds/month visibility before full Kubernetes adoption.
AI-written summary · about
chapters (23)
- 0:00Intro
- 1:17Gauravs background
- 3:16About Farfetch
- 7:18Brands we work with
- 7:41Size
- 8:46Customers
- 10:24How Do We Work
- 12:46Our Platform
- 15:36Born in the Cloud
- 17:56Live Infrastructure
- 21:58Simple Stack
- 26:42Trends
- 30:00Benefits of Kubernetes
- 33:05Prerequisites
- 37:53The Problem
- 38:19Blueprints
- 40:35Kafka
- 41:02Database dependencies
- 41:47Help charts in Kubernetes
- 42:06Observability
- 43:12Prometheus
- 43:49DayZero
- 44:22Testing Strategy
related talks
35mKubernetes journey
35mTo Infinite Scale and Beyond: Operating Kubernetes Past the Steady State
33mKubernetes-Driven Multi-Cloud Deployments: Maximizing Scalability and Flexibility
31mThe 20 Year Platform – bringing together Kubernetes, 12-Factor and Functions
53mDeploying to Kubernetes Thousands of Times Per/Day
46mKeep Building Fresh: Shopify's Journey to Kubernetes
39mStrangling the Legacy with Kubernetes
38mBuilding the Perfect Infrastructure with Kubernetes