SQL Group by & Max

Question

I have a following data in a table:

id  name    alarmId  alarmUnit  alarmLevel

1   test    voltage  psu        warning
2   test    voltage  psu        ceasing
3   test    voltage  psu        warning
4   test    temp     rcc        warning
5   test    temp     rcc        ceasing

I'd like to show only the most recent information about every colums group (alarmId,alarmUnit), so the result should look like this:

3   test    voltage  psu        warning
5   test    temp     rcc        ceasing

I've tried so far:

SELECT MAX(id) as id,name,alarmId,alarmUnit,alarmLevel GROUP BY alarmId,alarmUnit;

Selected IDs seem to be fine but selected rows aren't corresponding to them. Could you help me?

Answer 1

In Oracle, SQL Server 2005+ and PostgreSQL 8.4:

SELECT  *
FROM    (
        SELECT  *, ROW_NUMBER() OVER (PARTITION BY alarmId, alarmUnit ORDER BY id DESC) AS rn
        FROM    mytable
        ) q
WHERE   rn = 1

In MySQL:

SELECT  mi.*
FROM    (
        SELECT  alarmId, alarmUnit, MAX(id) AS mid
        FROM    mytable
        GROUP BY
                alarmId, alarmUnit
        ) mo
JOIN    mytable mi
ON      mi.id = mo.mid

In PostgreSQL 8.3 and below:

SELECT  DISTINCT ON (alarmId, alarmUnit) *
FROM    mytable
ORDER BY
        alarmId, alarmUnit, id DESC