Question

I have a table which goes like this

c_id           b_id
13              151
27              203
38              347
38              349
38              357
72              132
72              475
104             14
151             22
207             690
207             712

As you can see the table is ordered first by c_id and then by b_id. I would like to remove the first entry for every c_id.

So my output table should be this

c_id            b_id
38              349 
38              357
72              475
207             712

The c_ids which have just one b_id associated with them have been removed. And the first b_ids for the other c_ids have been removed.

Multiple queries can be used. There are millions of rows in the original table.

Answer 1

SELECT * 
FROM 
(
    SELECT c_id, b_id, 
        ROW_NUMBER() OVER(PARTITION BY c_id ORDER BY c_id DESC) RowNum
    FROM TestTable
    WHERE c_id IN 
       (
           SELECT c_id 
           FROM TestTable 
           GROUP BY c_id
           HAVING COUNT(b_id) > 1
       )
) t WHERE RowNum > 1

This should give you the following:

c_id   b_id   RowNum
--------------------
38     349     2
38     357     3
72     475     2
207    712     2

Live Demo