How do I limit the number of rows returned by an Oracle query after ordering?

Question

Is there a way to make an Oracle query behave like it contains a MySQL limit clause?

In MySQL, I can do this:

Answer 1

I'v started preparing for Oracle 1z0-047 exam, validated against 12c While prepping for it i came across a 12c enhancement known as 'FETCH FIRST' It enables you to fetch rows /limit rows as per your convenience. Several options are available with it

- FETCH FIRST n ROWS ONLY
 - OFFSET n ROWS FETCH NEXT N1 ROWS ONLY // leave the n rows and display next N1 rows
 - n % rows via FETCH FIRST N PERCENT ROWS ONLY

Example:

Select * from XYZ a
order by a.pqr
FETCH FIRST 10 ROWS ONLY

Answer 2

I did some performance testing for the following approaches:

Asktom

select * from (
  select a.*, ROWNUM rnum from (
    <select statement with order by clause>
  ) a where rownum <= MAX_ROW
) where rnum >= MIN_ROW

Analytical

select * from (
  <select statement with order by clause>
) where myrow between MIN_ROW and MAX_ROW

Short Alternative

select * from (
  select statement, rownum as RN with order by clause
) where a.rn >= MIN_ROW and a.rn <= MAX_ROW

Results

Table had 10 million records, sort was on an unindexed datetime row:

• Explain plan showed same value for all three selects (323168)
• But the winner is AskTom (with analytic following close behind)

Selecting first 10 rows took:

• AskTom: 28-30 seconds
• Analytical: 33-37 seconds
• Short alternative: 110-140 seconds

Selecting rows between 100,000 and 100,010:

• AskTom: 60 seconds
• Analytical: 100 seconds

Selecting rows between 9,000,000 and 9,000,010:

• AskTom: 130 seconds
• Analytical: 150 seconds

Answer 3

An analytic solution with only one nested query:

SELECT * FROM
(
   SELECT t.*, Row_Number() OVER (ORDER BY name) MyRow FROM sometable t
) 
WHERE MyRow BETWEEN 10 AND 20;

Rank() could be substituted for Row_Number() but might return more records than you are expecting if there are duplicate values for name.

Answer 4

If you are not on Oracle 12C, you can use TOP N query like below.

SELECT *
 FROM
   ( SELECT rownum rnum
          , a.*
       FROM sometable a 
   ORDER BY name
   )
WHERE rnum BETWEEN 10 AND 20;

You can even move this from clause in with clause as follows

WITH b AS
( SELECT rownum rnum
      , a.* 
   FROM sometable a ORDER BY name
) 
SELECT * FROM b 
WHERE rnum BETWEEN 10 AND 20;

Here actually we are creating a inline view and renaming rownum as rnum. You can use rnum in main query as filter criteria.

Answer 5

Pagination queries with ordering are really tricky in Oracle.

Oracle provides a ROWNUM pseudocolumn that returns a number indicating the order in which the database selects the row from a table or set of joined views.

ROWNUM is a pseudocolumn that gets many people into trouble. A ROWNUM value is not permanently assigned to a row (this is a common misunderstanding). It may be confusing when a ROWNUM value is actually assigned. A ROWNUM value is assigned to a row after it passes filter predicates of the query but before query aggregation or sorting.

What is more, a ROWNUM value is incremented only after it is assigned.

This is why the followin query returns no rows:

 select * 
 from (select *
       from some_table
       order by some_column)
 where ROWNUM <= 4 and ROWNUM > 1; 

The first row of the query result does not pass ROWNUM > 1 predicate, so ROWNUM does not increment to 2. For this reason, no ROWNUM value gets greater than 1, consequently, the query returns no rows.

Correctly defined query should look like this:

select *
from (select *, ROWNUM rnum
      from (select *
            from skijump_results
            order by points)
      where ROWNUM <= 4)
where rnum > 1; 

Find out more about pagination queries in my articles on Vertabelo blog:

• Oracle ROWNUM Explained
• Top-N and pagination queries

Answer 6

In oracle

SELECT val FROM   rownum_order_test ORDER BY val DESC FETCH FIRST 5 ROWS ONLY;

VAL

    10
    10
     9
     9
     8

5 rows selected.

SQL>