Scaling a Financial Institution with Kubernetes
summary
A financial company migrated 200+ services and millions of customers to Kubernetes by creating sharded clusters that each host a fixed set of clients with independent databases and replicas. They use git-backed service definitions plus a custom provisioner to generate cloud resources and Kubernetes objects; initial rollout tests on one shard revealed 5× latency spikes and resource rebalancing problems that were fixed, after which deployments scale better and recovery and development cycle times improve.
AI-written summary · about
chapters (9)
- 0:00Nubank introduction
- 2:09Sharding with Charlie
- 6:46Automation via definition
- 9:14Kubernetes challenges
- 10:59Kubernetes adoption benefits
- 13:29Kubernetes migration
- 16:16Engineer training
- 18:03Current Kubernetes setup
- 19:52Migration outcomes
related talks
47mScaling a financial institution with Kubernetes
25mBreaking Free From the Cloud: Banking on Self-Hosted Kubernetes
20mKeynote: The Challenges of Migrating 150+ Microservices to Kubernetes
35mKubernetes journey
52mDatabase sharding at the load balancer level
35mTo Infinite Scale and Beyond: Operating Kubernetes Past the Steady State
31mBeyond the Limits: Scaling Kubernetes Controllers Horizontally
38mBuilding the Perfect Infrastructure with Kubernetes