Question

I have a table; let it be called table1; with the following fields and data

alt text http://img228.imageshack.us/img228/3827/45939084.png

I need a query that returns the record with the maximum value in Field3 for each group of records having the same value in Field2. So that the query returns:

alt text http://img87.imageshack.us/img87/62/48847706.png

How could this be done using SQL queries ?

Answer 1

This:

WITH    q AS
        (
        SELECT  *, ROW_NUMBER() OVER (PARTITION BY field2 ORDER BY field3 DESC) AS rn
        FROM    table1
        )
SELECT  *
FROM    q
WHERE   rn = 1

or this:

SELECT  q.*
FROM    (
        SELECT  DISTINCT field2
        FROM    table1
        ) qo
CROSS APPLY
        (
        SELECT  TOP 1 *
        FROM    table1 t
        WHERE   t.field2 = qo.field2
        ORDER BY
                t.field3 DESC
        ) q

Depending on the field2 cardinality, the first or the second query can be more efficient.

See this article for more details:

• SQL Server: Selecting records holding group-wise maximum