How Iterable enables personalized experiences for billions of users
summary
Iterable uses CockroachDB since 2021 to serve user metadata for billions of people monthly and hundreds of millions daily with a under 10ms latency objective, running four clusters on EKS across three AZs via the operator pattern. Deployment tooling is Helm, Terraform and ArgoCD; monitoring with Datadog; cost savings by running on Graviton. Schemas optimize between hashed UUID keys to avoid hot ranges and native collocated keys for joins. Real‑time sync uses changefeed → S3 → Snowpipe, upgrades are rolling and largely painless.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 2:37CockroachDB at Iterable growth
- 3:42User identity lifecycle
- 6:31Scaling challenges
- 9:08Offloading to CockroachDB
- 10:30Optimizing latency
- 12:17Data modeling lessons
- 13:42Kubernetes setup
- 15:46Cluster management
- 17:09Lessons learned
- 18:13Key tooling
- 19:35Operator pain points
- 21:44Future plans
- 23:27Advice for developers
related talks
27mBeyond the Roadmap and Into the Future of CockroachDB
31mFuture-proofing operational data at scale
23mWhy Squarespace adopted CockroachDB for better global performance at scale
15mJourney from PostgreSQL to CockroachDB: Scaling Cisco AI Defense
48mBuilding mission-critical applications with CockroachDB
14mBuilding CockroachDB for performance at scale
18mFrom Today to 10X: Why Groww chose
32mHave a SQL and Eat it Too!