Data Engineering Books

Yes, as the second book rather than the first. Designing Data-Intensive Applications explains how the systems under a data platform work, from storage engines and encoding through replication and partitioning to transactions and batch and stream processing. Where Fundamentals tells you which decisions exist, this book tells you what each option costs when the data is large and the machines fail.

The book is now a 2nd edition. Martin Kleppmann and Chris Riccomini's edition was released by O'Reilly in March 2026, about 670 pages, and Kleppmann's announcement stresses cloud service design alongside the core principles of modern databases. If you own the 2017 first edition, its chapters on replication and consistency still hold, though the newer edition is the one to buy.

It is dense and rewards slow reading. Take 1 chapter at a time and write the failure case down: what happens to a write when a replica is partitioned away, or to a stream consumer that restarts. Those cases are the raw material of the data pipeline interview questions on delivery guarantees and on ordering during recovery, and you can try the design side on the data pipeline practice problems.

The Data Warehouse Toolkit by Ralph Kimball and Margy Ross, 3rd edition, published by Wiley in 2013. It is the reference for dimensional modeling: pick a business process, declare the grain, then attach the dimensions and the facts. Its case studies start with retail sales and inventory and go on to procurement and accounting, and the 3rd edition adds 2 chapters on ETL.

It is old, and it still matters. Warehouses have changed from appliances to cloud engines, but a data modeling round asks about the star schema and the surrogate key, and about slowly changing dimensions, and interviewers use Kimball's words. The ETL chapters are the part that has dated, since modern teams load first and transform in the warehouse.

Read the first 2 chapters and the case study closest to your own business, then come back for the rest as problems arise. To check yourself, see how dimensional modeling is asked in interviews, try the data modeling interview questions and work through the data modeling practice problems. For narrower drills there are dimensional modeling interview questions.

Data Engineering Design Patterns by Bartosz Konieczny (O'Reilly, April 2025) is the newest book here and the most practical. It's organised as recipes, with chapters on error management and idempotency and others on data value and data flow patterns. Read it when you're building or reviewing pipelines, because the ideas it names, such as replaying a partition safely, are the ones a design interview probes.

Streaming Systems by Tyler Akidau, Slava Chernyak and Reuven Lax (O'Reilly, 2018) is the best explanation of event time and of the windows and watermarks built on it, the ideas that make stream processing different from batch. The tools have moved since 2018, but the model has not. If you're weighing the 2 styles first, the guide to batch vs streaming pipelines frames the choice.

Kafka: The Definitive Guide, 2nd edition (November 2021), is by Gwen Shapira, Todd Palino, Rajini Sivaram and Krit Petty. Read it if Kafka is on your team's stack or your target company's. Learning Spark, 2nd edition, Spark 3.0 (July 2020), is by Jules Damji, Brooke Wenig, Tathagata Das and Denny Lee and teaches the structured APIs and how to read the Spark UI when a job is slow. Both were written for specific releases, so check the current release notes for anything newer than the book.

Concepts age slowly and products age fast, so sort each book by how much of it is which. Fundamentals of Data Engineering and The Data Warehouse Toolkit are almost all concepts. Designing Data-Intensive Applications is mostly concepts and was refreshed in 2026. Kafka: The Definitive Guide and Learning Spark were written for specific releases, and each predates the versions you will run today.

The common mistake is reading a tool book as if it were current. Learning Spark covers Spark 3.0, which is the right foundation for the structured APIs and the wrong place to learn about newer features. Use the book for the model, then the project's own documentation for the release you run.

Treat claims about editions with the same care. Some lists describe a new edition of Fundamentals of Data Engineering. The authors' own words are that a second edition is not ruled out and not imminent, so the June 2022 edition is the one to buy.

Books give you the reasoning, and an interview asks you to produce it under time pressure. Reading how a replica falls behind is different from explaining it while someone watches you draw. The books on this list cover the concept side of every round, and the rest is practice.

Match the book to the round. Fundamentals and the pattern book feed the system design round, Kimball feeds the data modeling round, and Kleppmann feeds the follow-up questions in both. SQL and Python rounds aren't covered by any of these books, so they need timed problems rather than reading. The data engineer interview prep guide lays out every round, and the data engineer roadmap places reading in the wider order of what to learn.

If you have 6 weeks, read Fundamentals in the first 2, take the Toolkit chapters you need in the next 2, and spend the last 2 on timed practice. Reading past that point returns less than solving a problem and checking your answer.