GopherCon 2022: Zach Musgrave - Performance in a High-throughput SQL Database

In this tutorial, Zach will show you how to profile your codebase to make it faster and share some performance-killing anti-patterns to avoid, all in the context of a novel SQL database written in golang.

Zach and the team at DoltHub built Dolt, the world’s first SQL database that you can branch and merge, fork and clone, push and pull like a git repository. They wrote the storage and query engine from the ground up, then made it as fast as MySQL by using golang’s profiler. Join Zach as he shares what they learned on this journey. He will demo how to use golang’s profiler tools to uncover slowdowns in your code. He will also detail many examples of slow code from DoltHub’s own code base and show you how the team fixed them. Finally, Zach will discuss the architectural choices for their database’s tuple storage, and how they contribute to its performance.