A NoSQL Deep Dive: Alex DeBrie on The DynamoDB Book

In the second half of our conversation with Alex DeBrie, we move past the fundamentals of his new book, The DynamoDB Book, to get his rapid-fire takes on everything from S3 pricing to the future of serverless databases.

S3 vs. DynamoDB for Storage Costs

DeBrie is clear that S3 remains the right choice for large blobs or data with a single access pattern. However, he points out a surprising nuance for those migrating heavy transactional workloads: write-heavy systems can actually end up costing more in S3 than in DynamoDB. The per-request pricing model flips the conventional wisdom for very active datasets.

GraphQL and the Case for Single-Table Design

Regarding the lack of support for single-table design in Amplify and GraphQL, DeBrie has made peace with the limitation. He argues that those services are optimizing for a different set of problems than bespoke single-table architectures. He remains a proponent of the Amplify team, describing them as one of the most unique groups at AWS, and wishes AWS would foster more of that contrarian thinking.

The Necessity of a 450-Page Database Guide

Defending the length of his book, DeBrie suggests that learning relational database management systems from scratch would likely require a similar volume of material. The difference is that RDBMS knowledge is distributed across decades of tutorials and coursework, whereas DynamoDB concepts are rarely taught incrementally in the same way.

Tooling: NoSQL Workbench

DeBrie is a fan of NoSQL Workbench for data modeling, visualization, and as living repository documentation. He jokes that it is a massive improvement over his previous Excel spreadsheet workflow, and he is bullish on the tool given the team's continued iteration.

Battle of the Serverless Databases: Aurora vs. DynamoDB

DeBrie admits his views have evolved since writing that Aurora Serverless was the future. Today, he views DynamoDB as the safer bet. He argues that learning DynamoDB's access patterns is more plausible than betting on achieving a fully serverless, pay-per-use, and infinitely scalable relational database.

The Relational Mindset

On the debate between hand-coding joins versus relying on materialized views, DeBrie stresses pragmatism. NoSQL patterns often appear odd to those from a relational world, and the difficulty lies in breaking that ingrained mindset. When asked if filtering on the client side is a database failure, he emphasizes context. The criticism forces developers to consider the problem holistically rather than forcing every operation into the database layer.

Holiday Wishes and Community Growth

DeBrie’s wish list for AWS Santa has yet to be fulfilled. Of his three major asks, he predicts only one—more Redis-like operations—might eventually land in DynamoDB. He also addresses the state of DynamoDB community tooling; while much of the developer experience is effectively outsourced to the community, he views this as a feature. Most initiatives will fail, he says, but the winners can be formalized into official support. DeBrie suggests that the influence of DynamoDB’s limitations is even visible in the design of the AWS console itself, noting that certain constraints have clearly shaped the user experience.