Is an index scan always faster than a sequential scan?

Definition:

An index scan follows an index to locate matching rows, while a sequential scan reads the table directly. If a query needs much of a table, following many scattered row locations can cost more than scanning it. PostgreSQL's planner weighs expected row count, I/O and other costs rather than always preferring indexes.

Key points:

• No. Reading much of a table may be faster with a sequential scan.

• Compare estimated rows, actual rows, filters and I/O.

• An index-only scan also depends on PostgreSQL visibility information, not just covering columns.

• An index-only scan can still visit the table to check visibility when visibility-map information is insufficient.

Interview point:

• A sequential scan is evidence to understand, not automatically a performance bug.