Columnar Storage
Storing table data by column instead of by row, so analytics that touch a few columns scan far less data.
Columnar storage physically stores a table’s data column-by-column rather than row-by-row. Analytical queries that read a few columns across many rows then scan only those columns, and similar values sitting together compress far better. It is why data warehouses are fast at aggregation.
Worked example: SELECT AVG(price) over a billion-row table reads just the price column in a columnar store, versus a row store that must read every full row; the same layout compresses that column heavily because its values are alike. Gotcha: columnar is great for scan-heavy analytics but poor for transactional writes — updating or inserting a single row touches every column file — which is exactly why OLTP databases stay row-oriented and OLAP warehouses go columnar.