Running total.

When OVER contains ORDER BY the aggregate becomes cumulative: SUM(total) OVER (ORDER BY order_date, id) adds every row from the start up to the current one. That is a "running balance" or sales so far over time. With PARTITION BY, accumulation restarts in each group.

Important: ORDER BY inside OVER belongs to the window only. It does NOT sort the final result — to see rows in order, add a separate ORDER BY at the end, usually with the same columns.

Equal values in OVER (ORDER BY order_date) are "peers": by default the window is RANGE from the start through the current row AND its peers, so two orders on the same day get the same running sum. For a row-by-row total add a unique key: ORDER BY order_date, id.

Common mistakes: OVER (ORDER BY order_date) with no tie-breaker; a missing final ORDER BY; believing OVER ORDER BY sorts the output; accumulating over a NULL customer when that is unwanted.

Small example:

SELECT id, order_date, total, SUM(total) OVER (ORDER BY order_date, id) AS running_total FROM orders ORDER BY order_date, id;