Sequential scan

The previous examples (and many in Chapter 9, Database Indexing) have shown you plenty of examples of tables being scanned sequentially. You can expect a Seq Scan when there isn't a useful index, or when such a large portion of the table is expected to be returned such that using an index would just add needless overhead. They'll also be used when there is only a very small amount of data to access; the index overhead is disproportionately large if the table only takes up a few pages on disk.

Note that a Seq Scan must read through all the dead rows in a table, but will not include them in its output. It's therefore possible for their execution to take much longer than it would be expected to produce all the required output, if the table has been badly maintained and is quite bloated with dead rows.