Why Row-Oriented Storage Wastes Your Database SSD

Lucas and Luna dive into the hidden inefficiency of row-oriented storage for modern analytics workloads. They explain how columnar storage flips the script by storing data by column instead of row, dramatically reducing I/O and improving query speed. The episode uses a concrete example: a 50-column sales table where a typical dashboard query touches only 5 columns. Lucas walks through what happens on disk with a row store versus a column store, citing 10x compression gains and query speedups of 50x or more for aggregation-heavy queries. They also discuss real-world trade-offs, like slower single-row lookups and the rise of hybrid formats like Apache Parquet. Luna challenges Lucas on whether this is just for data warehouses or if it applies to transactional databases too. The episode is lively, specific, and leaves listeners with a clear mental model: row stores read entire rows even when you need one column; column stores read only the columns you ask for. #DatabaseStorage #ColumnarStorage #RowOriented #DataEngineering #SQL #NoSQL #ApacheParquet #QueryPerformance #DataCompression #Analytics #IOOptimization #TechPodcast #DatabaseTech #FexingoBusiness #BusinessPodcast #Technology #DataWarehouse #LucasAndLuna Keep every episode free: buymeacoffee.com/fexingo