What's the difference between where clause and on clause when table left join?

Solution 1:

The where clause applies to the whole resultset; the on clause only applies to the join in question.

In the example supplied, all of the additional conditions related to fields on the inner side of the join - so in this example, the two queries are effectively identical.

However, if you had included a condition on a value in the table in the outer side of the join, it would have made a significant difference.

You can get more from this link: http://ask.sqlservercentral.com/questions/80067/sql-data-filter-condition-in-join-vs-where-clause

For example:

select t1.f1,t2.f2 from t1 left join t2 on t1.f1 = t2.f2 and t2.f4=1

select t1.f1,t2.f2 from t1 left join t2 on t1.f1 = t2.f2 where t2.f4=1

- do different things - the former will left join to t2 records where f4 is 1, while the latter has effectively been turned back into an inner join to t2.

Solution 2:

The first query is quicker than the second one as the join condition is more specific than the second one: it does not makes sense to return records that you will filter with the where clause (it would be better do not return them at all- query1)

Anyway it really depends by the query optimizer.

have a look at the below:

Is a JOIN faster than a WHERE?

Solution 3:

It is important to understand the logical order of SQL operations when thinking about SQL syntax. JOIN is an operator (and ON belongs to the relevant JOIN) in the FROM clause. The FROM clause is the first operation to be executed logically (optimisers can still choose to reorder things).

In your example, there isn't really a difference, but it is easy to construct one, as I've shown in this blog post about the difference between ON and WHERE in OUTER JOIN (the example from the blog post uses the Sakila database):

First query

SELECT a.actor_id, a.first_name, a.last_name, count(fa.film_id)
FROM actor a
LEFT JOIN film_actor fa ON a.actor_id = fa.actor_id
WHERE fa.film_id < 10
GROUP BY a.actor_id, a.first_name, a.last_name
ORDER BY count(fa.film_id) ASC;

Yields:

ACTOR_ID  FIRST_NAME  LAST_NAME  COUNT
--------------------------------------
194       MERYL       ALLEN      1
198       MARY        KEITEL     1
30        SANDRA      PECK       1
85        MINNIE      ZELLWEGER  1
123       JULIANNE    DENCH      1

Because we filtered the outer joined table in the WHERE clause, the LEFT JOIN was effectively turned into an INNER JOIN. Why? Because if we had an actor that didn't play in a film, that actor's only row would have fa.film_id IS NULL, and the fa.film_id < 10 predicate would thus yield NULL. Such actors are excluded from the result, just as with an INNER JOIN.

Second query

SELECT a.actor_id, a.first_name, a.last_name, count(fa.film_id)
FROM actor a
LEFT JOIN film_actor fa ON a.actor_id = fa.actor_id
  AND fa.film_id < 10
GROUP BY a.actor_id, a.first_name, a.last_name
ORDER BY count(fa.film_id) ASC;

Yields:

ACTOR_ID  FIRST_NAME  LAST_NAME     COUNT
-----------------------------------------
3         ED          CHASE         0
4         JENNIFER    DAVIS         0
5         JOHNNY      LOLLOBRIGIDA  0
6         BETTE       NICHOLSON     0
...
1         PENELOPE    GUINESS       1
200       THORA       TEMPLE        1
2         NICK        WAHLBERG      1
198       MARY        KEITEL        1

Now, the actors without films are included in the result, because the fa.film_id < 10 predicate is part of the LEFT JOIN's ON predicate

Conclusion

Always place predicates where they make most sense logically.

  • Are they part of your JOIN operation? Place them in ON
  • Are they filters on your entire JOIN product? Place them in WHERE

Solution 4:

The two queries are NOT identical.

Mark Bannister was right in pointing out that the where clause is applied to the whole result set but the on clause applies to the join.

In your case, for SQL 1 LEFT JOIN conditions filter joins on the right but the left side is always returned before any WHERE filtering. Since there are no WHERE conditions all of t1 is always returned.

In SQL 2, the LEFT JOIN conditions filter some results showing up on the right but again all t1 is returned. But this time the WHERE conditions may filter some records of t1 away.

INSERT INTO `t1` (`f1`,`f2`,`f3`) VALUES (1,1,1); INSERT INTO `t2` (`f3`) VALUES (1);

Since they point to different logic the query must be written based on that and it gives us great power and flexibility.

An INNER JOIN however returns the same result so yes check the optimiser.