5232

Also, how do LEFT OUTER JOIN, RIGHT OUTER JOIN, and FULL OUTER JOIN fit in?

4
  • 90
    Of the answers & comments & their references below only one actually explains how Venn diagrams represent the operators: The circle intersection area represents the set of rows in A JOIN B. The area unique to each circle represents the set of rows you get by taking its table's rows that don't participate in A JOIN B and adding the columns unique to the other table all set to NULL. (And most give a vague bogus correspondence of the circles to A and B.)
    – philipxy
    Oct 18, 2014 at 20:24
  • 22
    @DanteTheSmith No, that suffers from the same problems as the diagrams here. See my comment above re the question & below re that very blog post: "Jeff repudiates his blog a few pages down in the comments". Venn diagrams show elements in sets. Just try to identify exactly what the sets are and what the elements are in these diagrams. The sets aren't the tables and the elements aren't their rows. Also any two tables can be joined, so PKs & FKs are irrelvant. All bogus. You are doing just what thousands of others have done--got a vague impression you (wrongly) assume makes sense.
    – philipxy
    May 18, 2017 at 5:57
  • 2
    My preceding comment is about a confused repudiated Jeff Atwood blog post.
    – philipxy
    Jan 23, 2021 at 23:15
  • 2
    My 1st comment's link is external, but i.stack.imgur.com has permanent copies of its illustrations of output (not input) for inner, left & full joins (in green).
    – philipxy
    Feb 27, 2021 at 4:37

28 Answers 28

6702

Assuming you're joining on columns with no duplicates, which is a very common case:

  • An inner join of A and B gives the result of A intersect B, i.e. the inner part of a Venn diagram intersection.

  • An outer join of A and B gives the results of A union B, i.e. the outer parts of a Venn diagram union.

Examples

Suppose you have two tables, with a single column each, and data as follows:

A    B
-    -
1    3
2    4
3    5
4    6

Note that (1,2) are unique to A, (3,4) are common, and (5,6) are unique to B.

Inner join

An inner join using either of the equivalent queries gives the intersection of the two tables, i.e. the two rows they have in common.

select * from a INNER JOIN b on a.a = b.b;
select a.*, b.*  from a,b where a.a = b.b;

a | b
--+--
3 | 3
4 | 4

Left outer join

A left outer join will give all rows in A, plus any common rows in B.

select * from a LEFT OUTER JOIN b on a.a = b.b;
select a.*, b.*  from a,b where a.a = b.b(+);

a |  b
--+-----
1 | null
2 | null
3 |    3
4 |    4

Right outer join

A right outer join will give all rows in B, plus any common rows in A.

select * from a RIGHT OUTER JOIN b on a.a = b.b;
select a.*, b.*  from a,b where a.a(+) = b.b;

a    |  b
-----+----
3    |  3
4    |  4
null |  5
null |  6

Full outer join

A full outer join will give you the union of A and B, i.e. all the rows in A and all the rows in B. If something in A doesn't have a corresponding datum in B, then the B portion is null, and vice versa.

select * from a FULL OUTER JOIN b on a.a = b.b;

 a   |  b
-----+-----
   1 | null
   2 | null
   3 |    3
   4 |    4
null |    6
null |    5
13
  • 68
    It would be good to augment the example by adding another row in table B with value 4. This will show that inner joins need not be on equal no of rows.
    – softveda
    Aug 30, 2009 at 12:59
  • 537
    An excellent explanation, however this statement: An outer join of A and B gives the results of A union B, i.e. the outer parts of a venn diagram union. isn't phrased accurately. An outer join will give the results of A intersect B in addition to one of the following: all of A (left join), all of B (right join) or all of A and all of B (full join). Only this last scenario is really A union B. Still, a well written explanation.
    – Thomas
    May 3, 2011 at 19:57
  • 11
    @Ameer, Thanks. Join does not guarantee an order, you would need to add an ORDER BY clause. Mar 5, 2013 at 0:28
  • 10
    @Damian yes, OUTER JOIN and FULL OUTER JOIN are equivalent, along with LEFT/RIGHT JOIN are equivalent to LEFT/RIGHT OUTER JOIN, in the same way INNER JOIN is equivalent to a simple JOIN
    – Chris
    Jan 29, 2019 at 22:26
  • 13
    I have downvoted this because it is wrong. Please consider removing the answer as it will mislead generations of computer science students who are fooled by the large upcote count. Venn diagrams do not explain join. The inner part of a join is not intersection. May 3, 2020 at 7:00
962
+500

The Venn diagrams don't really do it for me.

They don't show any distinction between a cross join and an inner join, for example, or more generally show any distinction between different types of join predicate or provide a framework for reasoning about how they will operate.

There is no substitute for understanding the logical processing and it is relatively straightforward to grasp anyway.

  1. Imagine a cross join.
  2. Evaluate the on clause against all rows from step 1 keeping those where the predicate evaluates to true
  3. (For outer joins only) add back in any outer rows that were lost in step 2.

(NB: In practice the query optimiser may find more efficient ways of executing the query than the purely logical description above but the final result must be the same)

I'll start off with an animated version of a full outer join. Further explanation follows.

enter image description here


Explanation

Source Tables

enter link description here

First start with a CROSS JOIN (AKA Cartesian Product). This does not have an ON clause and simply returns every combination of rows from the two tables.

SELECT A.Colour, B.Colour FROM A CROSS JOIN B

enter link description here

Inner and Outer joins have an "ON" clause predicate.

  • Inner Join. Evaluate the condition in the "ON" clause for all rows in the cross join result. If true return the joined row. Otherwise discard it.
  • Left Outer Join. Same as inner join then for any rows in the left table that did not match anything output these with NULL values for the right table columns.
  • Right Outer Join. Same as inner join then for any rows in the right table that did not match anything output these with NULL values for the left table columns.
  • Full Outer Join. Same as inner join then preserve left non matched rows as in left outer join and right non matching rows as per right outer join.

Some examples

SELECT A.Colour, B.Colour FROM A INNER JOIN B ON A.Colour = B.Colour

The above is the classic equi join.

Inner Join

Animated Version

enter image description here

SELECT A.Colour, B.Colour FROM A INNER JOIN B ON A.Colour NOT IN ('Green','Blue')

The inner join condition need not necessarily be an equality condition and it need not reference columns from both (or even either) of the tables. Evaluating A.Colour NOT IN ('Green','Blue') on each row of the cross join returns.

inner 2

SELECT A.Colour, B.Colour FROM A INNER JOIN B ON 1 =1

The join condition evaluates to true for all rows in the cross join result so this is just the same as a cross join. I won't repeat the picture of the 16 rows again.

SELECT A.Colour, B.Colour FROM A LEFT OUTER JOIN B ON A.Colour = B.Colour

Outer Joins are logically evaluated in the same way as inner joins except that if a row from the left table (for a left join) does not join with any rows from the right hand table at all it is preserved in the result with NULL values for the right hand columns.

LOJ

SELECT A.Colour, B.Colour FROM A LEFT OUTER JOIN B ON A.Colour = B.Colour WHERE B.Colour IS NULL

This simply restricts the previous result to only return the rows where B.Colour IS NULL. In this particular case these will be the rows that were preserved as they had no match in the right hand table and the query returns the single red row not matched in table B. This is known as an anti semi join.

It is important to select a column for the IS NULL test that is either not nullable or for which the join condition ensures that any NULL values will be excluded in order for this pattern to work correctly and avoid just bringing back rows which happen to have a NULL value for that column in addition to the un matched rows.

loj is null

SELECT A.Colour, B.Colour FROM A RIGHT OUTER JOIN B ON A.Colour = B.Colour

Right outer joins act similarly to left outer joins except they preserve non matching rows from the right table and null extend the left hand columns.

ROJ

SELECT A.Colour, B.Colour FROM A FULL OUTER JOIN B ON A.Colour = B.Colour

Full outer joins combine the behaviour of left and right joins and preserve the non matching rows from both the left and the right tables.

FOJ

SELECT A.Colour, B.Colour FROM A FULL OUTER JOIN B ON 1 = 0

No rows in the cross join match the 1=0 predicate. All rows from both sides are preserved using normal outer join rules with NULL in the columns from the table on the other side.

FOJ 2

SELECT COALESCE(A.Colour, B.Colour) AS Colour FROM A FULL OUTER JOIN B ON 1 = 0

With a minor amend to the preceding query one could simulate a UNION ALL of the two tables.

UNION ALL

SELECT A.Colour, B.Colour FROM A LEFT OUTER JOIN B ON A.Colour = B.Colour WHERE B.Colour = 'Green'

Note that the WHERE clause (if present) logically runs after the join. One common error is to perform a left outer join and then include a WHERE clause with a condition on the right table that ends up excluding the non matching rows. The above ends up performing the outer join...

LOJ

... And then the "Where" clause runs. NULL= 'Green' does not evaluate to true so the row preserved by the outer join ends up discarded (along with the blue one) effectively converting the join back to an inner one.

LOJtoInner

If the intention was to include only rows from B where Colour is Green and all rows from A regardless the correct syntax would be

SELECT A.Colour, B.Colour FROM A LEFT OUTER JOIN B ON A.Colour = B.Colour AND B.Colour = 'Green'

enter image description here

SQL Fiddle

See these examples run live at SQLFiddle.com.

16
  • 69
    I will say that while this doesn't work for me nearly as well as the Venn diagrams, I appreciate that people vary and learn differently and this is a very well presented explanation unlike any I've seen before, so I support @ypercube in awarding the bonus points. Also good work explaining the difference of putting additional conditions in the JOIN clause vs the WHERE clause. Kudos to you, Martin Smith.
    – Old Pro
    Dec 20, 2014 at 4:48
  • 33
    @OldPro The Venn diagrams are OK as far as they go I suppose but they are silent on how to represent a cross join, or to differentiate one kind of join predicate such as equi join from another. The mental model of evaluating the join predicate on each row of the cross join result then adding back in unmatched rows if an outer join and finally evaluating the where works better for me. Dec 20, 2014 at 11:05
  • 24
    The Venn diagrams are good for representing Unions and Intersections and Differences but not joins. They have some minor educational value for very simple joins, i.e. joins where the joining condition is on unique columns. Dec 20, 2014 at 22:28
  • 17
    @Arth - Nope you're wrong. SQL Fiddle sqlfiddle.com/#!3/9eecb7db59d16c80417c72d1/5155 this is something the Venn diagrams can't illustrate. Jan 28, 2016 at 15:54
  • 9
    How did you do these animations? Great answer, the only bit I dislike is your modesty in saying that the Venn diagrams don't do it for you. The reality is that they are insufficient to model what's going on and this is important to tell, lest people get the wrong idea.
    – Git Gud
    May 23, 2018 at 17:55
252

Joins are used to combine the data from two tables, with the result being a new, temporary table. Joins are performed based on something called a predicate, which specifies the condition to use in order to perform a join. The difference between an inner join and an outer join is that an inner join will return only the rows that actually match based on the join predicate. For eg- Lets consider Employee and Location table:

Employee

EmpID EmpName
13 Jason
8 Alex
3 Ram
17 Babu
25 Johnson

Location

EmpID EmpLoc
13 San Jose
8 Los Angeles
3 Pune, India
17 Chennai, India
39 Bangalore, India

Inner Join:- Inner join creates a new result table by combining column values of two tables (Employee and Location) based upon the join-predicate. The query compares each row of Employee with each row of Location to find all pairs of rows which satisfy the join-predicate. When the join-predicate is satisfied by matching non-NULL values, column values for each matched pair of rows of Employee and Location are combined into a result row. Here’s what the SQL for an inner join will look like:

select  * from employee inner join location on employee.empID = location.empID
OR
select  * from employee, location where employee.empID = location.empID

Now, here is what the result of running that SQL would look like:

Employee.EmpId Employee.EmpName Location.EmpId Location.EmpLoc
13 Jason 13 San Jose
8 Alex 8 Los Angeles
3 Ram 3 Pune, India
17 Babu 17 Chennai, India

Outer Join:- An outer join does not require each record in the two joined tables to have a matching record. The joined table retains each record—even if no other matching record exists. Outer joins subdivide further into left outer joins and right outer joins, depending on which table's rows are retained (left or right).

Left Outer Join:- The result of a left outer join (or simply left join) for tables Employee and Location always contains all records of the "left" table (Employee), even if the join-condition does not find any matching record in the "right" table (Location). Here is what the SQL for a left outer join would look like, using the tables above:

select  * from employee left outer join location on employee.empID = location.empID;
//Use of outer keyword is optional

Now, here is what the result of running this SQL would look like:

Employee.EmpId Employee.EmpName Location.EmpId Location.EmpLoc
13 Jason 13 San Jose
8 Alex 8 Los Angeles
3 Ram 3 Pune, India
17 Babu 17 Chennai, India
25 Johnson NULL NULL

Note how while Johnson has no entry in the employee location table, he is still included in the results but the location fields are nulled.

Right Outer Join:- A right outer join (or right join) closely resembles a left outer join, except with the treatment of the tables reversed. Every row from the "right" table (Location) will appear in the joined table at least once. If no matching row from the "left" table (Employee) exists, NULL will appear in columns from Employee for those records that have no match in Location. This is what the SQL looks like:

select * from employee right outer join location  on employee.empID = location.empID;
//Use of outer keyword is optional

Using the tables above, we can show what the result set of a right outer join would look like:

Employee.EmpId Employee.EmpName Location.EmpId Location.EmpLoc
13 Jason 13 San Jose
8 Alex 8 Los Angeles
3 Ram 3 Pune, India
17 Babu 17 Chennai, India
NULL NULL 39 Bangalore, India

Note how while there are no employees listed as working in Bangalore, it is still included in the results with the employee fields nulled out.

Full Outer Joins:- Full Outer Join or Full Join is to retain the nonmatching information by including nonmatching rows in the results of a join, use a full outer join. It includes all rows from both tables, regardless of whether or not the other table has a matching value.

Employee.EmpId Employee.EmpName Location.EmpId Location.EmpLoc
13 Jason 13 San Jose
8 Alex 8 Los Angeles
3 Ram 3 Pune, India
17 Babu 17 Chennai, India
25 Johnson NULL NULL
NULL NULL 39 Bangalore, India

MySQL 8.3 Reference Manual - Join Syntax

Oracle Join operations

2
  • 1
    The Venn diagrams are mislabelled. See my comments on the question & other answers. Also most of this language is poor. Eg: "When the join-predicate is satisfied by matching non-NULL values, column values for each matched pair of rows of Employee and Location are combined into a result row." No, not "When the join-predicate is satisfied by matching non-NULL values". Values in rows don't matter other than whether the condition as a whole being true or false. Some values could well be NULL for a true condition.
    – philipxy
    Nov 6, 2019 at 7:00
  • @Persistence What is needed for the tables is text that is table initialization code formatted in columns that is suitable to copy & paste & run.
    – philipxy
    Nov 26, 2021 at 12:24
175

Inner Join

Retrieve the matched rows only, that is, A intersect B.

Enter image description here

SELECT *
FROM dbo.Students S
INNER JOIN dbo.Advisors A
    ON S.Advisor_ID = A.Advisor_ID

Left Outer Join

Select all records from the first table, and any records in the second table that match the joined keys.

Enter image description here

SELECT *
FROM dbo.Students S
LEFT JOIN dbo.Advisors A
    ON S.Advisor_ID = A.Advisor_ID

Full Outer Join

Select all records from the second table, and any records in the first table that match the joined keys.

Enter image description here

SELECT *
FROM dbo.Students S
FULL JOIN dbo.Advisors A
    ON S.Advisor_ID = A.Advisor_ID

References

6
  • 16
    What is the name of tool? I find it is interesting as it shows number of rows and venn-diagrams Jan 27, 2014 at 12:23
  • 3
    @GrijeshChauhan Yeah But you can Try to run it using wine . Jan 27, 2014 at 12:30
  • 2
    Ohh! yes I ..I used SQLyog using wine.. there is also PlayOnLinux Jan 27, 2014 at 12:32
  • 3
    Your text is unclear & wrong. The "matched rows only" are rows from the cross join of A & B & what is retrieved (A inner join B) is not A intersect B but (A left join B) intersect (A right join B). The "selected" rows are not from A & B, they are from A cross join B & from null-extended values of rows from A & B.
    – philipxy
    Jul 24, 2018 at 1:09
  • 1
    @TusharGupta-curioustushar you should include the "Tables used for SQL Examples" Oct 17, 2019 at 13:52
132

In simple words:

An inner join retrieve the matched rows only.

Whereas an outer join retrieve the matched rows from one table and all rows in other table ....the result depends on which one you are using:

  • Left: Matched rows in the right table and all rows in the left table

  • Right: Matched rows in the left table and all rows in the right table or

  • Full: All rows in all tables. It doesn't matter if there is a match or not

1
  • 2
    @nomen Not that this answer addresses it, but INNER JOIN is an intersection and FULL OUTER JOIN is the corresponding UNION if the left & right sets/circles contain the rows of (respectively) LEFT & RIGHT join. PS This answer is unclear about rows in input vs output. It confuses "in the left/right table" with "has a left/right part in the left/right" and it uses "matched row" vs "all" to mean row extended by row from other table vs by nulls.
    – philipxy
    Nov 29, 2015 at 1:17
128

A inner join only shows rows if there is a matching record on the other (right) side of the join.

A (left) outer join shows rows for each record on the left hand side, even if there are no matching rows on the other (right) side of the join. If there is no matching row, the columns for the other (right) side would show NULLs.

0
102

Inner joins require that a record with a related ID exist in the joined table.

Outer joins will return records for the left side even if nothing exists for the right side.

For instance, you have an Orders and an OrderDetails table. They are related by an "OrderID".

Orders

  • OrderID
  • CustomerName

OrderDetails

  • OrderDetailID
  • OrderID
  • ProductName
  • Qty
  • Price

The request

SELECT Orders.OrderID, Orders.CustomerName
  FROM Orders 
 INNER JOIN OrderDetails
    ON Orders.OrderID = OrderDetails.OrderID

will only return Orders that also have something in the OrderDetails table.

If you change it to OUTER LEFT JOIN

SELECT Orders.OrderID, Orders.CustomerName
  FROM Orders 
  LEFT JOIN OrderDetails
    ON Orders.OrderID = OrderDetails.OrderID

then it will return records from the Orders table even if they have no OrderDetails records.

You can use this to find Orders that do not have any OrderDetails indicating a possible orphaned order by adding a where clause like WHERE OrderDetails.OrderID IS NULL.

1
  • 1
    I appreciate the simple yet realistic example. I changed a request like SELECT c.id, c.status, cd.name, c.parent_id, cd.description, c.image FROM categories c, categories_description cd WHERE c.id = cd.categories_id AND c.status = 1 AND cd.language_id = 2 ORDER BY c.parent_id ASC to SELECT c.id, c.status, cd.name, c.parent_id, cd.description, c.image FROM categories c INNER JOIN categories_description cd ON c.id = cd.categories_id WHERE c.status = 1 AND cd.language_id = 2 ORDER BY c.parent_id ASC (MySQL) with success. I wasn't sure about the additional conditions, they mix well...
    – PhiLho
    Jan 5, 2013 at 11:11
85

In simple words :

Inner join -> Take ONLY common records from parent and child tables WHERE primary key of Parent table matches Foreign key in Child table.

Left join ->

pseudo code

1.Take All records from left Table
2.for(each record in right table,) {
    if(Records from left & right table matching on primary & foreign key){
       use their values as it is as result of join at the right side for 2nd table.
    } else {
       put value NULL values in that particular record as result of join at the right side for 2nd table.
    }
  }

Right join : Exactly opposite of left join . Put name of table in LEFT JOIN at right side in Right join , you get same output as LEFT JOIN.

Outer join : Show all records in Both tables No matter what. If records in Left table are not matching to right table based on Primary , Forieign key , use NULL value as result of join .

Example :

Example

Lets assume now for 2 tables

1.employees , 2.phone_numbers_employees

employees : id , name 

phone_numbers_employees : id , phone_num , emp_id   

Here , employees table is Master table , phone_numbers_employees is child table(it contains emp_id as foreign key which connects employee.id so its child table.)

Inner joins

Take the records of 2 tables ONLY IF Primary key of employees table(its id) matches Foreign key of Child table phone_numbers_employees(emp_id).

So query would be :

SELECT e.id , e.name , p.phone_num FROM employees AS e INNER JOIN phone_numbers_employees AS p ON e.id = p.emp_id;

Here take only matching rows on primary key = foreign key as explained above.Here non matching rows on primary key = foreign key are skipped as result of join.

Left joins :

Left join retains all rows of the left table, regardless of whether there is a row that matches on the right table.

SELECT e.id , e.name , p.phone_num FROM employees AS e LEFT JOIN phone_numbers_employees AS p ON e.id = p.emp_id;

Outer joins :

SELECT e.id , e.name , p.phone_num FROM employees AS e OUTER JOIN phone_numbers_employees AS p ON e.id = p.emp_id;

Diagramatically it looks like :

Diagram

3
  • 5
    The result has nothing to (do per se) with primary/unique/candidate keys & foreign keys. The baviour can and should be described without reference to them. A cross join is calculated, then rows not matching the ON condition are filtered out; additionally for outer joins rows filtered/unmatched rows are extended by NULLs (per LEFT/RIGHT/FULL and included.
    – philipxy
    Aug 10, 2015 at 4:27
  • 1
    The assumption that SQL joins are always a match on primary/foreign keys is leading to this misuse of Venn diagrams. Please revise your answer accordingly. May 7, 2020 at 18:37
  • Such Venn-like diagrams for joins are unclear, unhelpful & misleading. See my Q&A comments here. "All you have to do to see this is try to write a correct legend for one." Don't forget that SQL tables are bags not sets. See my comments on the question & answers & see my answer & its quotes & links.
    – philipxy
    Mar 15 at 3:05
70

INNER JOIN requires there is at least a match in comparing the two tables. For example, table A and table B which implies A ٨ B (A intersection B).

LEFT OUTER JOIN and LEFT JOIN are the same. It gives all the records matching in both tables and all possibilities of the left table.

Similarly, RIGHT OUTER JOIN and RIGHT JOIN are the same. It gives all the records matching in both tables and all possibilities of the right table.

FULL JOIN is the combination of LEFT OUTER JOIN and RIGHT OUTER JOIN without duplication.

70

You use INNER JOIN to return all rows from both tables where there is a match. i.e. In the resulting table all the rows and columns will have values.

In OUTER JOIN the resulting table may have empty columns. Outer join may be either LEFT or RIGHT.

LEFT OUTER JOIN returns all the rows from the first table, even if there are no matches in the second table.

RIGHT OUTER JOIN returns all the rows from the second table, even if there are no matches in the first table.

0
56

The answer is in the meaning of each one, so in the results.

Note :
In SQLite there is no RIGHT OUTER JOIN or FULL OUTER JOIN.
And also in MySQL there is no FULL OUTER JOIN.

My answer is based on above Note.

When you have two tables like these:

--[table1]               --[table2]
id | name                id | name
---+-------              ---+-------
1  | a1                  1  | a2
2  | b1                  3  | b2

CROSS JOIN / OUTER JOIN :
You can have all of those tables data with CROSS JOIN or just with , like this:

SELECT * FROM table1, table2
--[OR]
SELECT * FROM table1 CROSS JOIN table2

--[Results:]
id | name | id | name 
---+------+----+------
1  | a1   | 1  | a2
1  | a1   | 3  | b2
2  | b1   | 1  | a2
2  | b1   | 3  | b2

INNER JOIN :
When you want to add a filter to above results based on a relation like table1.id = table2.id you can use INNER JOIN:

SELECT * FROM table1, table2 WHERE table1.id = table2.id
--[OR]
SELECT * FROM table1 INNER JOIN table2 ON table1.id = table2.id

--[Results:]
id | name | id | name 
---+------+----+------
1  | a1   | 1  | a2

LEFT [OUTER] JOIN :
When you want to have all rows of one of tables in the above result -with same relation- you can use LEFT JOIN:
(For RIGHT JOIN just change place of tables)

SELECT * FROM table1, table2 WHERE table1.id = table2.id 
UNION ALL
SELECT *, Null, Null FROM table1 WHERE Not table1.id In (SELECT id FROM table2)
--[OR]
SELECT * FROM table1 LEFT JOIN table2 ON table1.id = table2.id

--[Results:]
id | name | id   | name 
---+------+------+------
1  | a1   | 1    | a2
2  | b1   | Null | Null

FULL OUTER JOIN :
When you also want to have all rows of the other table in your results you can use FULL OUTER JOIN:

SELECT * FROM table1, table2 WHERE table1.id = table2.id
UNION ALL
SELECT *, Null, Null FROM table1 WHERE Not table1.id In (SELECT id FROM table2)
UNION ALL
SELECT Null, Null, * FROM table2 WHERE Not table2.id In (SELECT id FROM table1)
--[OR] (recommended for SQLite)
SELECT * FROM table1 LEFT JOIN table2 ON table1.id = table2.id
UNION ALL
SELECT * FROM table2 LEFT JOIN table1 ON table2.id = table1.id
WHERE table1.id IS NULL
--[OR]
SELECT * FROM table1 FULL OUTER JOIN table2 On table1.id = table2.id

--[Results:]
id   | name | id   | name 
-----+------+------+------
1    | a1   | 1    | a2
2    | b1   | Null | Null
Null | Null | 3    | b2

Well, as your need you choose each one that covers your need ;).

0
45

Inner join.

A join is combining the rows from two tables. An inner join attempts to match up the two tables based on the criteria you specify in the query, and only returns the rows that match. If a row from the first table in the join matches two rows in the second table, then two rows will be returned in the results. If there’s a row in the first table that doesn’t match a row in the second, it’s not returned; likewise, if there’s a row in the second table that doesn’t match a row in the first, it’s not returned.

Outer Join.

A left join attempts to find match up the rows from the first table to rows in the second table. If it can’t find a match, it will return the columns from the first table and leave the columns from the second table blank (null).

37

enter image description here

  • INNER JOIN most typical join for two or more tables. It returns data match on both table ON primarykey and forignkey relation.
  • OUTER JOIN is same as INNER JOIN, but it also include NULL data on ResultSet.
    • LEFT JOIN = INNER JOIN + Unmatched data of left table with Null match on right table.
    • RIGHT JOIN = INNER JOIN + Unmatched data of right table with Null match on left table.
    • FULL JOIN = INNER JOIN + Unmatched data on both right and left tables with Null matches.
  • Self join is not a keyword in SQL, when a table references data in itself knows as self join. Using INNER JOIN and OUTER JOIN we can write self join queries.

For example:

SELECT * 
FROM   tablea a 
       INNER JOIN tableb b 
               ON a.primary_key = b.foreign_key 
       INNER JOIN tablec c 
               ON b.primary_key = c.foreign_key 
0
34

I don't see much details about performance and optimizer in the other answers.

Sometimes it is good to know that only INNER JOIN is associative which means the optimizer has the most option to play with it. It can reorder the join order to make it faster keeping the same result. The optimizer can use the most join modes.

Generally it is a good practice to try to use INNER JOIN instead of the different kind of joins. (Of course if it is possible considering the expected result set.)

There are a couple of good examples and explanation here about this strange associative behavior:

3
  • 5
    It can't possibly be "good practice" to use one type of join over another. Which join you use determines the data that you want. If you use a different one you're incorrect. Plus, in Oracle at least this answer is completely wrong. It sounds completely wrong for everything and you have no proof. Do you have proof?
    – Ben
    Dec 26, 2014 at 8:51
  • 1
    1. I mean try to use. I saw lots of people using LEFT OUTER joins everywhere without any good reason. (The joined columns were 'not null'.) In those cases it would be definitely better to use INNER joins. 2. I have added a link explaining the non-associative behaviour better than I could. Dec 26, 2014 at 11:01
  • As I know INNER JOIN is slower than LEFT JOIN in most of the times, And people can use LEFT JOIN instead of INNER JOIN by adding a WHERE for removing unexpected NULL results ;).
    – shA.t
    May 13, 2015 at 3:01
33

Having criticized the much-loved red-shaded Venn diagram, I thought it only fair to post my own attempt.

Although @Martin Smith's answer is the best of this bunch by a long way, his only shows the key column from each table, whereas I think ideally non-key columns should also be shown.

The best I could do in the half hour allowed, I still don't think it adequately shows that the nulls are there due to absence of key values in TableB or that OUTER JOIN is actually a union rather than a join:

enter image description here

2
  • 2
    Question is asking for Difference between INNER and OUTER joins though, not necessarily left outer join Jan 22, 2016 at 19:32
  • @LearnByReading: my picture on the right is a right outer join i.e. replace TableA a LEFT OUTER JOIN TableB b with TableB B RIGHT OUTER JOIN TableA a
    – onedaywhen
    Apr 29, 2019 at 7:58
33

The precise algorithm for INNER JOIN, LEFT/RIGHT OUTER JOIN are as following:

  1. Take each row from the first table: a
  2. Consider all rows from second table beside it: (a, b[i])
  3. Evaluate the ON ... clause against each pair: ON( a, b[i] ) = true/false?
    • When the condition evaluates to true, return that combined row (a, b[i]).
    • When reach end of second table without any match, and this is an Outer Join then return a (virtual) pair using Null for all columns of other table: (a, Null) for LEFT outer join or (Null, b) for RIGHT outer join. This is to ensure all rows of first table exists in final results.

Note: the condition specified in ON clause could be anything, it is not required to use Primary Keys (and you don't need to always refer to Columns from both tables)! For example:

Inner Join vs. Left Outer Join


enter image description here

Note: Left Join = Left Outer Join, Right Join = Right Outer Join.

1
  • How does matching work for "ON T1.col1 = T2.col2 + 1"? What actually gets matched between the two columns?
    – wenn32
    Sep 7, 2023 at 8:04
33

The General Idea

Please see the answer by Martin Smith for a better illustations and explanations of the different joins, including and especially differences between FULL OUTER JOIN, RIGHT OUTER JOIN and LEFT OUTER JOIN.

These two table form a basis for the representation of the JOINs below:

Basis

CROSS JOIN

CrossJoin

SELECT *
  FROM citizen
 CROSS JOIN postalcode

The result will be the Cartesian products of all combinations. No JOIN condition required:

CrossJoinResult

INNER JOIN

INNER JOIN is the same as simply: JOIN

InnerJoin

SELECT *
  FROM citizen    c
  JOIN postalcode p ON c.postal = p.postal

The result will be combinations that satisfies the required JOIN condition:

InnerJoinResult

LEFT OUTER JOIN

LEFT OUTER JOIN is the same as LEFT JOIN

LeftJoin

SELECT *
  FROM citizen         c
  LEFT JOIN postalcode p ON c.postal = p.postal

The result will be everything from citizen even if there are no matches in postalcode. Again a JOIN condition is required:

LeftJoinResult

Data for playing

All examples have been run on an Oracle 18c. They're available at dbfiddle.uk which is also where screenshots of tables came from.

CREATE TABLE citizen (id      NUMBER,
                      name    VARCHAR2(20),
                      postal  NUMBER,  -- <-- could do with a redesign to postalcode.id instead.
                      leader  NUMBER);

CREATE TABLE postalcode (id      NUMBER,
                         postal  NUMBER,
                         city    VARCHAR2(20),
                         area    VARCHAR2(20));

INSERT INTO citizen (id, name, postal, leader)
              SELECT 1, 'Smith', 2200,  null FROM DUAL
        UNION SELECT 2, 'Green', 31006, 1    FROM DUAL
        UNION SELECT 3, 'Jensen', 623,  1    FROM DUAL;

INSERT INTO postalcode (id, postal, city, area)
                 SELECT 1, 2200,     'BigCity',         'Geancy'  FROM DUAL
           UNION SELECT 2, 31006,    'SmallTown',       'Snizkim' FROM DUAL
           UNION SELECT 3, 31006,    'Settlement',      'Moon'    FROM DUAL  -- <-- Uuh-uhh.
           UNION SELECT 4, 78567390, 'LookoutTowerX89', 'Space'   FROM DUAL;

Blurry boundaries when playing with JOIN and WHERE

CROSS JOIN

CROSS JOIN resulting in rows as The General Idea/INNER JOIN:

SELECT *
  FROM citizen          c
  CROSS JOIN postalcode p
 WHERE c.postal = p.postal -- < -- The WHERE condition is limiting the resulting rows

Using CROSS JOIN to get the result of a LEFT OUTER JOIN requires tricks like adding in a NULL row. It's omitted.

INNER JOIN

INNER JOIN becomes a cartesian products. It's the same as The General Idea/CROSS JOIN:

SELECT *
  FROM citizen    c
  JOIN postalcode p ON 1 = 1  -- < -- The ON condition makes it a CROSS JOIN

This is where the inner join can really be seen as the cross join with results not matching the condition removed. Here none of the resulting rows are removed.

Using INNER JOIN to get the result of a LEFT OUTER JOIN also requires tricks. It's omitted.

LEFT OUTER JOIN

LEFT JOIN results in rows as The General Idea/CROSS JOIN:

SELECT *
  FROM citizen         c
  LEFT JOIN postalcode p ON 1 = 1 -- < -- The ON condition makes it a CROSS JOIN

LEFT JOIN results in rows as The General Idea/INNER JOIN:

SELECT *
  FROM citizen         c
  LEFT JOIN postalcode p ON c.postal = p.postal
 WHERE p.postal IS NOT NULL -- < -- removed the row where there's no mathcing result from postalcode

The troubles with the Venn diagram

An image internet search on "sql join cross inner outer" will show a multitude of Venn diagrams. I used to have a printed copy of one on my desk. But there are issues with the representation.

Venn diagram are excellent for set theory, where an element can be in one or both sets. But for databases, an element in one "set" seem, to me, to be a row in a table, and therefore not also present in any other tables. There is no such thing as one row present in multiple tables. A row is unique to the table.

Self joins are a corner case where each element is in fact the same in both sets. But it's still not free of any of the issues below.

The set A represents the set on the left (the citizen table) and the set B is the set on the right (the postalcode table) in below discussion.

CROSS JOIN

Every element in both sets are matched with every element in the other set, meaning we need A amount of every B elements and B amount of every A elements to properly represent this Cartesian product. Set theory isn't made for multiple identical elements in a set, so I find Venn diagrams to properly represent it impractical/impossible. It doesn't seem that UNION fits at all.

The rows are distinct. The UNION is 7 rows in total. But they're incompatible for a common SQL results set. And this is not how a CROSS JOIN works at all:

CrossJoinUnion1

Trying to represent it like this:

CrossJoinUnion2Crossing

..but now it just looks like an INTERSECTION, which it's certainly not. Furthermore there's no element in the INTERSECTION that is actually in any of the two distinct sets. However, it looks very much like the searchable results similar to this:

CrossJoinUnionUnion3

For reference one searchable result for CROSS JOINs can be seen at Tutorialgateway. The INTERSECTION, just like this one, is empty.

INNER JOIN

The value of an element depends on the JOIN condition. It's possible to represent this under the condition that every row becomes unique to that condition. Meaning id=x is only true for one row. Once a row in table A (citizen) matches multiple rows in table B (postalcode) under the JOIN condition, the result has the same problems as the CROSS JOIN: The row needs to be represented multiple times, and the set theory isn't really made for that. Under the condition of uniqueness, the diagram could work though, but keep in mind that the JOIN condition determines the placement of an element in the diagram. Looking only at the values of the JOIN condition with the rest of the row just along for the ride:

InnerJoinIntersection - Filled

This representation falls completely apart when using an INNER JOIN with a ON 1 = 1 condition making it into a CROSS JOIN.

With a self-JOIN, the rows are in fact idential elements in both tables, but representing the tables as both A and B isn't very suitable. For example a common self-JOIN condition that makes an element in A to be matching a different element in B is ON A.parent = B.child, making the match from A to B on seperate elements. From the examples that would be a SQL like this:

SELECT *
  FROM citizen c1
  JOIN citizen c2 ON c1.id = c2.leader

SelfJoinResult

Meaning Smith is the leader of both Green and Jensen.

OUTER JOIN

Again the troubles begin when one row has multiple matches to rows in the other table. This is further complicated because the OUTER JOIN can be though of as to match the empty set. But in set theory the union of any set C and an empty set, is always just C. The empty set adds nothing. The representation of this LEFT OUTER JOIN is usually just showing all of A to illustrate that rows in A are selected regardless of whether there is a match or not from B. The "matching elements" however has the same problems as the illustration above. They depend on the condition. And the empty set seems to have wandered over to A:

LeftJoinIntersection - Filled

WHERE clause - making sense

Finding all rows from a CROSS JOIN with Smith and postalcode on the Moon:

SELECT *
  FROM citizen          c
 CROSS JOIN postalcode  p
 WHERE c.name = 'Smith'
   AND p.area = 'Moon';

Where - result

Now the Venn diagram isn't used to reflect the JOIN. It's used only for the WHERE clause:

Where

..and that makes sense.

When INTERSECT and UNION makes sense

INTERSECT

As explained an INNER JOIN is not really an INTERSECT. However INTERSECTs can be used on results of seperate queries. Here a Venn diagram makes sense, because the elements from the seperate queries are in fact rows that either belonging to just one of the results or both. Intersect will obviously only return results where the row is present in both queries. This SQL will result in the same row as the one above WHERE, and the Venn diagram will also be the same:

SELECT *
  FROM citizen          c
 CROSS JOIN postalcode  p
 WHERE c.name = 'Smith'
INTERSECT
SELECT *
  FROM citizen          c
 CROSS JOIN postalcode  p
 WHERE p.area = 'Moon';

UNION

An OUTER JOIN is not a UNION. However UNION work under the same conditions as INTERSECT, resulting in a return of all results combining both SELECTs:

SELECT *
  FROM citizen          c
 CROSS JOIN postalcode  p
 WHERE c.name = 'Smith'
UNION
SELECT *
  FROM citizen          c
 CROSS JOIN postalcode  p
 WHERE p.area = 'Moon';

which is equivalent to:

SELECT *
  FROM citizen          c
 CROSS JOIN postalcode  p
 WHERE c.name = 'Smith'
   OR p.area = 'Moon';

..and gives the result:

Union - Result

Also here a Venn diagram makes sense:

UNION

When it doesn't apply

An important note is that these only work when the structure of the results from the two SELECT's are the same, enabling a comparison or union. The results of these two will not enable that:

SELECT *
  FROM citizen
 WHERE name = 'Smith'
SELECT *
  FROM postalcode
 WHERE area = 'Moon';

..trying to combine the results with UNION gives a

ORA-01790: expression must have same datatype as corresponding expression

For further interest read Say NO to Venn Diagrams When Explaining JOINs and sql joins as venn diagram. Both also cover EXCEPT.

0
25

Simplest Definitions

Inner Join: Returns matched records from both tables.

Full Outer Join: Returns matched and unmatched records from both tables with null for unmatched records from Both Tables.

Left Outer Join: Returns matched and unmatched records only from table on Left Side.

Right Outer Join: Returns matched and unmatched records only from table on Right Side.

In-Short

Matched + Left Unmatched + Right Unmatched = Full Outer Join

Matched + Left Unmatched = Left Outer Join

Matched + Right Unmatched = Right Outer Join

Matched = Inner Join

2
  • 1
    This is brilliant and explains why join doesn't work as expected for Time Series index's. Time stamps one second apart are unmatched. Sep 5, 2017 at 9:22
  • 1
    @yeliabsalohcin You don't explain "as expected" here or "works" in your comment on the question. It's just some unexplained personal misconception you strangely expect others to have. If you treat words as sloppily when you are reading--misinterpreting clear writing and/or accepting unclear writing--as when you are writing here then you can expect to have misconceptions. In fact this answer like most here is unclear & wrong. "Inner Join: Returns matched records from both tables" is wrong when input column sets differ. It's trying to say a certain something, but it isn't. (See my answer.)
    – philipxy
    Nov 22, 2017 at 2:52
14

In Simple Terms,

1.INNER JOIN OR EQUI JOIN : Returns the resultset that matches only the condition in both the tables.

2.OUTER JOIN : Returns the resultset of all the values from both the tables even if there is condition match or not.

3.LEFT JOIN : Returns the resultset of all the values from left table and only rows that match the condition in right table.

4.RIGHT JOIN : Returns the resultset of all the values from right table and only rows that match the condition in left table.

5.FULL JOIN : Full Join and Full outer Join are same.

10

left join on returns inner join on rows union all unmatched left table rows extended by nulls.

right join on returns inner join on rows union all unmatched right table rows extended by nulls.

full join on returns inner join on rowsunion all unmatched left table rows extended by nulls union all unmatched right table rows extended by nulls.

outer is optional & has no effect.

(SQL Standard 2006 SQL/Foundation 7.7 Syntax Rules 1, General Rules 1 b, 3 c & d, 5 b.)

So don't outer join on until you know what underlying inner join on is involved.


Find out what rows inner join on returns: CROSS JOIN vs INNER JOIN in SQL

That also explains why Venn(-like) diagrams are not helpful for inner vs outer join. For more on why they are not helpful for joins generally: Venn Diagram for Natural Join

5
  • I have indeed read your many comments. When you say, "a Venn diagram, when properly interpreted, can represent inner vs outer join" do you mean when properly interpreted by the observer or the Venn diagram itself? If the latter, please draw it :)
    – onedaywhen
    Apr 24, 2019 at 16:08
  • 1
    I'm not sure what you are trying to say. I am talking about the standard interpretation of a Venn diagram as sets of elements. (Because some uses of diagrams don't even manage that.) "Properly" for an application includes saying what the sets and/or elements are. See comment at the top of this page with 50 upvotes re a Venn diagram for inner vs outer joins. I'll edit some of my comments into this question. I don't want a Venn diagram in this post.
    – philipxy
    Apr 24, 2019 at 16:31
  • I must admit that, despite my quick phrasing in comments, because SQL involves bags & nulls and SQL culture doesn't have common terminology to name & distinguish between relevant notions, it is non-trivial even to explain clearly how elements of a Venn diagram are 1:1 with output "rows", let alone input "rows". Or what inner or outer joins do, let alone their difference. "value" may or may not include NULL, "row" may be a list of values vs a slot in a table value or variable & "=" may be SQL "=" vs equality.
    – philipxy
    Apr 25, 2019 at 17:44
  • Similar to our Cartesian-product-vs-relational-product discussion, I suspect it is the case that the Venn diagrams make a lot of sense to folk who already understand the differences between the join types!
    – onedaywhen
    Apr 29, 2019 at 7:51
  • In the case of 'relational Cartesian product', that is a standard & reasonable name for a certain thing that people do generally understand & which is reasonably described as similar to a Cartesian product. In the case of SQL Venn diagrams, they don't make sense, people just assume they do, whether they do or don't understand the operators/differences.
    – philipxy
    Apr 29, 2019 at 11:15
5

Joins are more easily explained with an example:

enter image description here

To simulate persons and emails stored in separate tables,

Table A and Table B are joined by Table_A.id = Table_B.name_id

Inner Join

enter image description here

Only matched ids' rows are shown.

Outer Joins

enter image description here

Matched ids and not matched rows for Table A are shown.

enter image description here

Matched ids and not matched rows for Table B are shown.

enter image description here Matched ids and not matched rows from both Tables are shown.

Note: Full outer join is not available on MySQL

4

1.Inner Join: Also called as Join. It returns the rows present in both the Left table, and right table only if there is a match. Otherwise, it returns zero records.

Example:

SELECT
  e1.emp_name,
  e2.emp_salary    
FROM emp1 e1
INNER JOIN emp2 e2
  ON e1.emp_id = e2.emp_id

output1

2.Full Outer Join: Also called as Full Join. It returns all the rows present in both the Left table, and right table.

Example:

SELECT
  e1.emp_name,
  e2.emp_salary    
FROM emp1 e1
FULL OUTER JOIN emp2 e2
  ON e1.emp_id = e2.emp_id

output2

3.Left Outer join: Or simply called as Left Join. It returns all the rows present in the Left table and matching rows from the right table (if any).

4.Right Outer Join: Also called as Right Join. It returns matching rows from the left table (if any), and all the rows present in the Right table.

joins

Advantages of Joins

  1. Executes faster.
0
4

Consider below 2 tables:

EMP

empid   name    dept_id salary
1       Rob     1       100
2       Mark    1       300
3       John    2       100
4       Mary    2       300
5       Bill    3       700
6       Jose    6       400

Department

deptid  name
1       IT
2       Accounts
3       Security
4       HR
5       R&D

Inner Join:

Mostly written as just JOIN in sql queries. It returns only the matching records between the tables.

Find out all employees and their department names:

Select a.empid, a.name, b.name as dept_name
FROM emp a
JOIN department b
ON a.dept_id = b.deptid
;

empid   name    dept_name
1       Rob     IT
2       Mark    IT
3       John    Accounts
4       Mary    Accounts
5       Bill    Security

As you see above, Jose is not printed from EMP in the output as it's dept_id 6 does not find a match in the Department table. Similarly, HR and R&D rows are not printed from Department table as they didn't find a match in the Emp table.

So, INNER JOIN or just JOIN, returns only matching rows.

LEFT JOIN :

This returns all records from the LEFT table and only matching records from the RIGHT table.

Select a.empid, a.name, b.name as dept_name
FROM emp a
LEFT JOIN department b
ON a.dept_id = b.deptid
;

empid   name    dept_name
1       Rob     IT
2       Mark    IT
3       John    Accounts
4       Mary    Accounts
5       Bill    Security
6       Jose    

So, if you observe the above output, all records from the LEFT table(Emp) are printed with just matching records from RIGHT table.

HR and R&D rows are not printed from Department table as they didn't find a match in the Emp table on dept_id.

So, LEFT JOIN returns ALL rows from Left table and only matching rows from RIGHT table.

Can also check DEMO here.

0
4

There are a lot of good answers here with very accurate relational algebra examples. Here is a very simplified answer that might be helpful for amateur or novice coders with SQL coding dilemmas.

Basically, more often than not, JOIN queries boil down to two cases:

For a SELECT of a subset of A data:

  • use INNER JOIN when the related B data you are looking for MUST exist per database design;
  • use LEFT JOIN when the related B data you are looking for MIGHT or MIGHT NOT exist per database design.
3
  • Inner join - An inner join using either of the equivalent queries gives the intersection of the two tables, i.e. the two rows they have in common.

  • Left outer join - A left outer join will give all rows in A, plus any common rows in B.

  • Full outer join - A full outer join will give you the union of A and B, i.e. All the rows in A and all the rows in B. If something in A doesn't have a corresponding datum in B, then the B portion is null, and vice versay

4
  • 2
    This is both wrong and unclear. Join is not an intersection unless the tables have the same columns. Outer joins don't have rows from A or B unless they have the same columns, in which case there are not nulls added. You are trying to say something, but you are not saying it. You are not explaining correctly or clearly.
    – philipxy
    Feb 3, 2017 at 11:15
  • @philipxy: Disagreed on your statement Join is not an intersection unless the tables have the same columns No. You can join any columns that you want and if the value match, they will join together. Apr 11, 2017 at 9:23
  • 1
    That comment is as unclear as your answer. (I suppose you might be thinking something like, the set of subrow values for the common columns of the result is the intersection of the sets of subrow values for the common columns of each of the inputs; but that's not what you have written. You are not clear.)
    – philipxy
    Apr 11, 2017 at 14:24
  • 1
    What I meant was that join is only an intersection of inputs when it is a natural inner join of inputs with the same columns. You are using the words "intersection" & "union" wrongly.
    – philipxy
    Feb 5, 2020 at 12:34
3

The difference between inner join and outer join is as follow:

  1. Inner join is a join that combined tables based on matching tuples, whereas outer join is a join that combined table based on both matched and unmatched tuple.
  2. Inner join merges matched row from two table in where unmatched row are omitted, whereas outer join merges rows from two tables and unmatched rows fill with null value.
  3. Inner join is like an intersection operation, whereas outer join is like an union operation.
  4. Inner join is two types, whereas outer join are three types.
  5. outer join is faster than inner join.
2
  • 2
    An outer join result is the same as inner join but plus some additional rows so I have no idea why you think outer join would be faster. Also what are these "two types" of inner join? I suppose you are referring to full,left, and right for outer? Jan 9, 2020 at 14:07
  • 1
    Outer join is not faster than inner join.
    – philipxy
    Jan 9, 2020 at 14:10
3

The "outer" and "inner" are just optional elements, you are just dealing with two (three) kinds of joins. Inner joins (or what is the default when using only "join") is a join where only the elements that match the criteria are present on both tables.

The "outer" joins are the same as the inner join plus the elements of the left or right table that didn't match, adding nulls on all columns for the other table.

The full join is the inner plus the right and left joins.

In summary, if we have table A like this

idA ColumnTableA idB
1 Jonh 1
2 Sarah 1
3 Clark 2
4 Barbie NULL

And table B like this:

idB ColumnTableB
1 Connor
2 Kent
3 Spock

The inner join:

from tableA join tableB on tableA.idB = tableB.idB
idA ColumnTableA idB ColumnTableB
1 Jonh 1 Connor
2 Sarah 1 Connor
3 Clark 2 Kent

Left outer join:

from tableA left join tableB on tableA.idB = tableB.idB
idA ColumnTableA idB ColumnTableB
1 Jonh 1 Connor
2 Sarah 1 Connor
3 Clark 2 Kent
4 Barbie NULL NULL

Right outer join:

from tableA right join tableB on tableA.idB = tableB.idB
idA ColumnTableA idB ColumnTableB
1 Jonh 1 Connor
2 Sarah 1 Connor
3 Clark 2 Kent
NULL NULL 3 Spock

Full outer join:

from tableA full join tableB on tableA.idB = tableB.idB
idA ColumnTableA idB ColumnTableB
1 Jonh 1 Connor
2 Sarah 1 Connor
3 Clark 2 Kent
4 Barbie NULL NULL
NULL NULL 3 Spock
2

A Demonstration

Setup

Hop into psql and create a tiny database of cats and humans. You can just copy-paste this whole section.

CREATE DATABASE catdb;
\c catdb;
\pset null '[NULL]' -- how to display null values

CREATE TABLE humans (
  name text primary key
);
CREATE TABLE cats (
  human_name text references humans(name),
  name text
);

INSERT INTO humans (name)
VALUES ('Abe'), ('Ann'), ('Ben'), ('Jen');

INSERT INTO cats (human_name, name)
VALUES
('Abe', 'Axel'),
(NULL, 'Bitty'),
('Jen', 'Jellybean'),
('Jen', 'Juniper');

Querying

Here's a query we'll run several times, changing [SOMETHING JOIN] to the various types to see the results.

SELECT
humans.name AS human_name,
cats.name AS cat_name
FROM humans
[SOMETHING JOIN] cats ON humans.name = cats.human_name
ORDER BY humans.name;

An INNER JOIN returns all human-cat pairs. Any human without a cat or cat without a human is excluded.

 human_name | cat_name
------------+-----------
 Abe        | Axel
 Jen        | Jellybean
 Jen        | Juniper

A FULL OUTER JOIN returns all humans and all cats, with NULL if there is no match on either side.

 human_name | cat_name
------------+-----------
 Abe        | Axel
 Ann        | [NULL]
 Ben        | [NULL]
 Jen        | Jellybean
 Jen        | Juniper
 [NULL]     | Bitty

A LEFT OUTER JOIN returns all humans (the left table). Any human without a cat gets a NULL in the cat_name column. Any cat without a human is excluded.

 human_name | cat_name
------------+-----------
 Abe        | Axel
 Ann        | [NULL]
 Ben        | [NULL]
 Jen        | Jellybean
 Jen        | Juniper

A RIGHT OUTER JOIN returns all cats (the right table). Any cat without a human gets a NULL in the human_name column. Any human without a cat is excluded.

 human_name | cat_name
------------+-----------
 Abe        | Axel
 Jen        | Jellybean
 Jen        | Juniper
 [NULL]     | Bitty

INNER vs OUTER

You can see that while an INNER JOIN gets only matching pairs, each kind of OUTER join includes some items without a match.

However, the actual words INNER and OUTER do not need to appear in queries:

  • JOIN by itself implies INNER
  • LEFT JOIN, RIGHT JOIN and OUTER JOIN all imply OUTER
0

Not the answer you're looking for? Browse other questions tagged or ask your own question.