Turning Databases Inside Out: A Paradigm Shift

Martin Kleppmann's talk challenges the conventional database architecture. He proposes a revolutionary approach: inverting the database. Instead of the traditional global, shared, mutable state, Kleppmann suggests viewing a database as an ever-growing collection of immutable facts. Using a distributed stream processing framework like Apache Samza, data streams are processed in real-time. At its core is a distributed, durable commit log (e.g., Apache Kafka). This approach promises simpler code, better scalability and robustness, lower latency, and greater flexibility for data manipulation.