The ROWID Scan algorithm is used to locate efficiently a row in a base or temporary table based on an equality comparison predicate that uses the ROWID function. The comparison predicate may refer to a constant literal, but more commonly the ROWID function is used with a row identifier value returned by a system function or procedure call, such as sa_locks.
ROWID Scans appear in the short and long plan as
<correlation_name><rowID>, where correlation_name is the correlation name specified in the FROM clause, or the table name if no correlation name was specified.
It is impossible for the ROWID Scan algorithm to differentiate between an invalid row identifier for the given table referenced by the ROWID function, and a situation where the given row identifier no longer exists. Consequently, the algorithm returns the empty set if the row identifier specified in the comparison predicate cannot be found in the table.