Secondary Indexes

Welcome to the eighth chapter of The Definitive Guide! By now, you’ve probably developed a good understanding of the principles and data patterns that can be used to model data in DynamoDB. These include considerations such as when to denormalize data to avoid making runtime compute-intensive joins, when to duplicate data to prebuild the data as your application would need it, and when to go for a multi-table strategy as opposed to a single-table one. There is obviously more to do with NoSQL and DynamoDB than just these concepts, but in a nutshell, they do represent some of the key aspects.

When DynamoDB was launched back in 2012 (1), it was quickly adopted by customers. This was also documented only a few months later by Amazon’s CTO, Werner Vogels, in a blog post (2) where he publicly shared how the data stored by customers in DynamoDB was doubling every couple of months and the cumulative throughput provisioned on the tables by these customers...