dynotable

Make sure to prewarm your table with WarmThroughput if you need to support more than 4k writes per second

DynamoDB cannot run JOINs and aggregations natievly, usually it is solved by exporting to Athena (for big datasets) or writing a one-off script.

We took a different approach and built a local SQL engine on top of SQLite that can run JOINs, SUM, AVG, MIN, MAX and automatically selects best GSI for efficient queries.

It is still Scans and Queries under the hood and is limitet to joins over hash and range keys, but it efficiently replaces one off scripts and simplifies life of people (and AI agents) who are used to SQL and it gives familiar abstraction where you need to think less about how it works and focus on the result.

89M rps was the peak across all Amazon systems during 2021 Prime Day. Alexa, Amazon.com and the fulfilment centres together, not one site. But its still very impressieve. But the more importarnt number is the per partition limit, 3000 RCU and 1000 WCU. Horizontal scale is the easy part, a table only gets near it if the partition key actually spreads.