Get rowns that don't exist in any of the tables.

Question

Hi,
I need to implement some views to compare systems.
This views will compare tables with the same structure and I want to extract the rows that don’t exist in any of the sides.
I need to implement some views to compare only two tables, others to compare 3 tables and one to compare 5 tables and another to compare 10 tables!
Each table can contains from a few rows up to 10 million rows (that was the biggest count I found for this tables).
My test scenario:
CREATE TABLE TEST_TABLE
(COL1 NUMBER ,
COL2 CHAR(25));
CREATE TABLE TEST_TABLE2
(COL1 NUMBER ,
COL2 CHAR(25));
CREATE TABLE TEST_TABLE3
(COL1 NUMBER ,
COL2 CHAR(25));
insert into TEST_TABLE (COL1,COL2) values (1,'2');
insert into TEST_TABLE (COL1,COL2) values (11,'t1');
insert into TEST_TABLE2 (COL1,COL2)  values (1,'2');
insert into TEST_TABLE2 (COL1,COL2) values (22,'t2');
insert into TEST_TABLE3 (COL1,COL2) values (1,'2');
insert into TEST_TABLE3 (COL1,COL2) values (33,'t3');To find the differences between two tables I implemented the following:
select * from
(select * from TEST_TABLE
minus
select * from TEST_TABLE2
union all
(select * from TEST_TABLE2
minus
select * from TEST_TABLE
Result:
COL1 COL2
11     t1                      
22     t2                       For the 3 tables comparison The result should be:
Result:
COL1 COL2
11     t1                      
22     t2                      
33     t3                       For the other ones I can implement the same way, but for sure this is not the prettiest and most performing solution!
How can I achieve the result I intent with the most performance?
Thanks,
Ricardo Tomás

Answer 1

You didn't say if you allow duplicates in your tables, so i assumed you didn't.
If a single table could have multiple occurrences of a given col1, col2 combination you would need to distinct the col1,col2 list from each table before doing the union all.
ME_XE?  select
  2     col1, col2
  3  from
  4  (
  5     select col1, col2       from test_table
  6             union all
  7     select col1, col2       from test_table2
  8             union all
  9     select col1, col2       from test_table3
10  )
11  group by
12     col1, col2
13  having count(*) = 1;
              COL1 COL2
                33 t3
                22 t2
                11 t1
3 rows selected.
Elapsed: 00:00:00.01
ME_XE?