Running Fast SQL on DynamoDB Tables

Using these features, Rockset continuously ingests data from DynamoDB in two steps:

To ensure Rockset does not lose any new data which is recorded in the DynamoDB table when the scan is happening, Rockset enables strongly consistent scans in the Rockset-DynamoDB connector, and also creates DynamoDB Streams (if not already present) and records the sequence numbers of existing shards. Primary key values from the DynamoDB table are used to construct the field in Rockset to uniquely identify a document in a Rockset collection. For this example, I have created a DynamoDB table programmatically using a Hacker News data set. In this entire process, I simply created a Rockset collection with a DynamoDB source, without any data transformation and schema modeling, and immediately ran SQL queries over it. Using Rockset, you too can join data across different DynamoDB tables or other sources to power your live applications.

Source: rockset.com