Database Internals

About the book

Database Internals is Alex Petrov's technical book on the systems that store and retrieve data at scale. It divides its attention between storage engines, the local machinery of writing and reading records, and distributed systems, the coordination problems that appear once data lives on many machines. The book assumes comfort with systems programming and treats its subject as engineering with trade-offs.

Petrov works from the mechanics outward. Storage topics include on-disk structures, file formats, and the buffer management that sits between disk and memory, alongside the indexing and retrieval strategies that determine how quickly a query can find what it wants. The distributed half turns to replication, partitioning, and consensus, the recurring problems of keeping copies consistent and agreeing on order when machines fail or messages arrive late.

The approach is comparative. The book sets competing structures and protocols beside each other and asks what each one costs. That framing suits readers who want to reason about a new engine or a production incident instead of memorizing one vendor's documentation.

How to read it

Read the storage engine section first and keep a pencil nearby. The material rewards slow attention to diagrams and to the difference between structures that look similar on paper but behave differently under writes, reads, and recovery. Pause on the trade-off discussions; they carry more of the book's value than any single definition.

The distributed systems half is denser and assumes some familiarity with networking and concurrency. Expect to reread the consensus material, and treat the protocol descriptions as arguments about failure modes rather than recipes to implement. The book is long and technical. Readers who prefer narrative will find it dry, while those who like precise mechanism will find the pace steady.

Who it suits

Software engineers who build or operate storage systems, backend developers who want to understand why a database behaves as it does, and students moving from general systems courses toward database engineering will get the most from it. It suits readers who already know basic data structures and want the next layer of detail about persistence, indexing, and coordination.

Reading considerations

This is a demanding technical work, not an entry point to databases, and readers without systems background will struggle with the distributed chapters. Some product references and implementation details may have moved on; check current documentation before relying on specifics. Test the comparative claims against a system you know rather than accepting them as settled.