Amazon DynamoDB: A Scalable, Predictably Performant, and Fully Managed NoSQL...
summary
Amazon reports DynamoDB serves trillions of API calls and peaks at 89.2 million requests/sec during Prime Day while keeping single‑digit millisecond latencies and regional 4‑nines / global 5‑nines SLAs. Operational fixes include token‑bucket bursting, a global admission control layer to allow cross‑partition bursts, and a horizontally scalable memory data store (MDS) using a Patricia+Merkle hybrid for partition lookups; these changes stop cache‑thundering and restore availability after a 400× metadata storm.
AI-written summary · about
chapters (14)
- 0:00Introduction
- 0:25Fully Managed NoSQL
- 2:52DynamoDB Stats
- 3:44Dynamo vs DynamoDB
- 4:29DynamoDB Architecture
- 5:33Why DynamoDB
- 6:46Partitioning
- 9:09Admission Control
- 10:53Overprovisioning
- 12:36Uniform Distribution
- 13:03Global Admission Control
- 14:27Partition Burst
- 18:47Lessons Learned
- 19:49Conclusion
related talks
30mBuild for Scale, Performance With Feature-packed Amazon DynamoDB
33mBuild for scale with Amazon DynamoDB developer features
29mPurpose-Built Databases for Modern Applications
29mPurpose-Built Databases for Modern Applications
59mModernize your applications with purpose-built databases
21mSave, Grow and Innovate using AWS Databases
24mBuild modern applications with purpose-built AWS databases
20mMaximise database availability