Question

For example given this table

id | category

1 | 10

2 | 10

3 | 10

4 | 10

5 | 20

6 | 20

7 | 20

8 | 30

9 | 30

10 | 30

11 | 30

We want obtain first 2 small id, from each category, that is we need this result

1

2

5

6

8

9

I wriite this query and his works

        SELECT MIN(id) AS id FROM mytable GROUP BY category  
        UNION 
        SELECT MIN(id) AS id FROM mytable WHERE 
        id NOT IN (SELECT MIN(id) AS id FROM mytable GROUP BY category) GROUP BY category

        ORDER BY id

but here is one problem, this works if we want first 2 id from each category, but if we want more first small ids (for example 7) from each category, query will be very difficult.
Someone has idea, how can make this easy?

Answer 1

@Quassnoi wrote a couple of very informative blog articles on this subject, concluding that the following would be extremely performant if you have a composite index on (category, id):

SELECT mytable.*
FROM   mytable JOIN (
  SELECT category, (
    SELECT   id
    FROM     mytable
    WHERE    mytable.category = categories.category
    ORDER BY id
    LIMIT    1, 1
  ) AS id
  FROM (
    SELECT DISTINCT category
    FROM   mytable
  ) AS categories
) AS limits ON mytable.category  <= limits.category
           AND mytable.category  >= limits.category
           AND IFNULL(mytable.id <= limits.id, TRUE)

See it on sqlfiddle.

To select the first 4 records, change LIMIT 1,1 to LIMIT 3,1. More generally: to select the first n records, change LIMIT 1,1 to LIMIT n-1,1.