Explore the development of Khepri, a tree-like, replicated database library created by the RabbitMQ team to replace Mnesia for persistent and replicated metadata storage. In this 20-minute conference talk from Code BEAM Europe 2023, Michael Davis delves into the reasons behind Khepri's creation, provides basic usage examples and implementation details, and discusses recent significant changes. Learn about the new "projections" feature that aims to match or surpass Mnesia's read speed in many scenarios. Discover how the API has been overhauled for improved ergonomics and Elixir compatibility. Get updates on Khepri's integration progress into the RabbitMQ codebase, including plans to introduce it in the next minor release under a feature flag. Gain insights into the future of distributed database management in RabbitMQ and the potential impact of Khepri on system performance and reliability.