Amazon DynamoDB with Flo Health: Powering Data for 300M Users
summary
Flo Health runs user profiles for 300 million users on Amazon DynamoDB, using auto scaling, GSIs and encryption to hit single‑digit ms latency; notifications use a TTL‑driven scheduler but TTL lacks delivery SLA so an extra notification trigger compensates; record size constraints (400 KB) force vertical partitioning to cut write RCUs and costs; production footprint shows 20 shards × 30M items and the setup scales and stays available while reducing cost.
AI-written summary · about
chapters (15)
- 0:00Introduction to DynamoDB at Flo Health
- 1:27What is Flo Health
- 5:06DynamoDB in Flo’s architecture
- 8:57Personalized notifications with TTL
- 13:57Privacy and security in DynamoDB
- 17:42Six years of DynamoDB at Flo
- 18:03DynamoDB key features overview
- 20:54DynamoDB terminology explained
- 23:26Global secondary indexes in DynamoDB
- 25:11Capacity modes: provisioned vs on-demand
- 28:01Data distribution and partitioning
- 31:50When to use on-demand capacity
- 33:58Cost optimization techniques
- 37:36Vertical partitioning for cost savings
- 41:54Summary and key takeaways
related talks
20mAmazon DynamoDB: A Scalable, Predictably Performant, and Fully Managed NoSQL...
30mBuild for Scale, Performance With Feature-packed Amazon DynamoDB
33mBuild for scale with Amazon DynamoDB developer features
39mBuild Your Data Lake on AWS — The Flo Experience
21mSave, Grow and Innovate using AWS Databases
33mDynamoDB Essentials and Integration for Java Developers
29mPurpose-Built Databases for Modern Applications
37mDynamoDB in an EDA World