Come back to any tab,
right on time.

How B-trees keep a database fast

Every time you look something up in a database, a B-tree is probably doing the work. It keeps millions of keys in order and finds any one of them in a handful of steps, even when most of the data lives on disk.

The trick is that each node holds many keys, not two. A node is sized to fit one disk page, so a single read brings in hundreds of keys at once, and the tree stays only three or four levels deep.

When a node fills up, it splits in two and passes its middle key up to its parent. That one rule keeps the tree balanced without ever rebuilding it.