Adding column in table having millions of records

Question

Hi,
Oracle release :11.2.0.3
Table A exists with millions of records. - has many indexes on it.
Need to add one column which exists in table B with same no of records.
Rowid is common join condition between table A and B to update this new column in A.
Please advice the fastest way to update this column.
Explain plan output for update query is :
| Id  | Operation                                                   | Name                  | Rows  | Bytes | Cost (%CPU)| Time     |
|   0 | UPDATE STATEMENT                             |                       |   181M|  3287M|    95T  (1)|999:59:59 |
|   1 |  UPDATE                                                | A                     |       |       |            |          |
|   2 |   PX COORDINATOR                                |                       |       |       |            |          |
|   3 |    PX SEND QC (RANDOM)                       | :TQ10000              |   181M|  3287M|   205K  (2)| 00:47:58 |
|   4 |     PX BLOCK ITERATOR                           |                       |   181M|  3287M|   205K  (2)| 00:47:58 |
|   5 |      TABLE ACCESS FULL                         | A                 |   181M|  3287M|   205K  (2)| 00:47:58 |
PLAN_TABLE_OUTPUT
|*  6 |   FILTER                                                    |                       |       |       |            |          |
|*  7 |    TABLE ACCESS BY INDEX ROWID          | B        |   301K|    15M|   528K  (1)| 02:03:24 |
|*  8 |     INDEX RANGE SCAN                              | SYS_C0073404          |    30M|       | 31081   (1)| 00:07:16 |
Thanks in advance

Answer 1

create table new_A as select *.A ,column_in_B from A,B where A.row_id=B.row_id;
drop table A;
rename new_A to A;
No - that can't be right.
You need to access the ROWID of table A; not some column in table A that is named 'row_id'. And that assumes, as PaulHorth asked, that table B DOES have a column named 'row_id' that contains the ROWID values of table A.
Also you only posted the plan for the update statement: post the actual query that plan is based on.
create table emp_rowid as
select 'new_' || e.ename new_ename, e.rowid row_id from emp e
where deptno = 20
update /*+parallel (4) */ emp e
set ename = (select new_ename from emp_rowid er
         where er.row_id = e.rowid)
where rowid in (select row_id from emp_rowid)        
select * from table(dbms_xplan.display_cursor())