Same Job, Two Engines: InnoDB and Iceberg Internals

A developer migrated their stock analysis system from InnoDB (OLTP) to an Iceberg-based OLAP stack after discovering that OLTP engines with their row-oriented storage and heavy indexing were inefficient for analytical workloads. The article explains how InnoDB's design—featuring row-based storage, multiple B-tree indexes, and MVCC for concurrent access—is optimized for transactional operations but poorly suited for analytics, leading to a painful redesign after the database grew to over 100GB. The switch to OLAP architecture better suited the analytical query patterns, demonstrating how choosing the right database engine architecture is crucial for workload optimization.