Similar to Oracle's implementation, IBM's DB2 has an internally generated unique row identifier called a Rowid. The implementation of this, however, is different than Oracle's. To use a Rowid in DB2, a column must be created on a table using the Rowid datatype. This column can then be retrieved in a Select and stored in a program's host variable. It can later be used to directly retrieve the row in the table to perform a Select, Update, or Delete. When using a Rowid in this way, a row is directly accessed without performing a Table scan and without needing to use an Index. Here is an example:

Select * from table_name where rowid_column = :hostVarRowid;

Notice that in DB2, Rowid is actually a datatype of a column.