Balancing Power and Pain: Moving a Startup From a PaaS to Kubernetes
Talkat KubeCon + CloudNativeCon North America 2019 · on Thursday, 2019-11-21· at 10:55· Room 14AB - San Diego Convention Center, Mezzanine Level·
summary
A three‑person ops effort moves a Heroku-hosted Rails stack onto GKE with Terraform, CircleCI, Harness, Ambassador, Prometheus, Jaeger and Sumo Logic, fixing Heroku limits (10 dynos, ~14.5GB RAM) and a Postgres shared_buffers outage that lasted four days. Migration starts as an October 2018 hackathon, launched production April 2019, then the monolith follows six weeks later. Results: saved $70k/month (from ~$110k to ~$40k), solved cross‑AZ latency by choosing a closer Google region, and containerized tricky deps like wkhtmltopdf.
AI-written summary · about
chapters (17)
- 0:00Intro
- 0:36Who are you
- 1:24About Gusto
- 3:01Heroku
- 4:57Scale
- 7:48Two Phases
- 8:55Define Requirements
- 9:11QA Engineers
- 10:34Platform Team
- 11:13Overwhelming Ecosystem
- 12:17Hackathon
- 13:15Challenges
- 18:37Unforeseen Challenges
- 21:32Amazon Web Services Gravity
- 23:23Continuous Delivery Pipeline
- 29:39Reflections
- 32:44Questions
related talks
24mFrom Lift-and-Shift to K8s: A Two-Phase DevOps Modernization
28mMoving a Monolith from AWS EC2 to K8
44mMigrating a Monolith to Kubernetes
28mSimplifying at Scale: 7 Years of Rails Architecture
35mKubernetes journey
20mKeynote: The Challenges of Migrating 150+ Microservices to Kubernetes
38mThe Reality of Porting Kubernetes Apps Across Cloud Providers
35mThe Oregon Trail to Kubernetes [I]