Join (SQL)

From Free net encyclopedia

A join combines records from two or more tables in a relational database. In the Structured Query Language (SQL), there are three types of joins: inner, outer, and cross. Outer joins are subdivided further into left outer joins, right outer joins, and full outer joins.

Contents 1 Example Tables 2 Join forms 2.1 Cross joins 2.2 Inner join 2.3 Left outer join 2.4 Right outer join 2.5 Full outer join 3 Implementation 3.1 Join optimization 3.2 Join algorithms 3.2.1 Nested loops 3.2.2 Merge join 3.2.3 Hash join 3.2.4 Semi join 4 See also 5 External links

Example Tables

The join examples below use the following two tables:

Table "employee" LastName DepartmentID Smith 34 Jones 33 Robinson 34 Jasper 36 Steinberg 33 Rafferty 31

Table "department" DepartmentName DepartmentID Sales 31 Engineering 33 Clerical 34 Marketing 35

Join forms

Cross joins

While not used very commonly, a cross join is the foundation upon which inner joins are built. A cross join returns the cartesian product of the sets of rows from the joined tables.

The SQL code for a cross join lists the tables to be joined (FROM), but does not include any filtering predicate (WHERE).

Example cross join (ANSI 92 standard syntax):

SELECT *
FROM employee CROSS JOIN 
     department;

Example cross join (non-standard syntax):

SELECT *
FROM employee
    ,department;

Example results:

LastName	DepartmentID	DepartmentName	DepartmentID
Smith	34	Sales	31
Smith	34	Engineering	33
Smith	34	Clerical	34
Smith	34	Marketing	35
Jones	33	Sales	31
Jones	33	Engineering	33
Jones	33	Clerical	34
Jones	33	Marketing	35
...16 more rows...

As you can see the cross join does no matching of like records. These joins are occasionally used to generate all possible combinations of records from tables that do not share a common element.

Inner join

An inner join essentially finds the intersection between the two tables. This is the most common type of join used, and is considered the default join type. The join example below takes all the records from table A (in this case, employee) and finds the matching record(s) from table B (department). If no match is found, the record from A is not included in the results. If multiple results are found in B that match the predicate then one row will be returned for each (the values from A will be repeated).

Special care must be taken when joining tables on columns that can be NULL since NULL values will never match each other. See Left Outer Join or Right Outer Join for a solution.

Example inner join (ANSI 92 standard syntax):

SELECT *  
FROM employee 
     INNER JOIN 
     department 
       ON employee.DepartmentID = department.DepartmentID

Example inner join (non-standard syntax):

SELECT *  
FROM employee
    ,department 
WHERE employee.DepartmentID = department.DepartmentID

LastName	DepartmentID	DepartmentName	DepartmentID
Smith	34	Clerical	34
Jones	33	Engineering	33
Robinson	34	Clerical	34
Steinberg	33	Engineering	33
Rafferty	31	Sales	31

Left outer join

A left outer join is very different from an inner join. Instead of limiting results to those in both tables, it limits results to those in the "left" table (A). This means that if the ON clause matches 0 records in B, a row in the result will still be returned—but with NULL values for each column from B.

For example, this allows us to find the employee's departments, but still show the employee even when their department is NULL or does not exist. The example above would have ignored employees in non-existent departments.

Example left outer join (ANSI 92 standard syntax):

SELECT *  
FROM employee 
     LEFT OUTER JOIN 
     department 
       ON employee.DepartmentID = department.DepartmentID

LastName	DepartmentID	DepartmentName	DepartmentID
Smith	34	Clerical	34
Jones	33	Engineering	33
Robinson	34	Clerical	34
Jasper	36	NULL	NULL
Steinberg	33	Engineering	33
Rafferty	31	Sales	31

Right outer join

A right outer join is much like a left outer join, except that the tables are reversed. Every record from the right side, B or department, will be returned, and NULL values will be returned for those that have no matching record in A.

Example right outer join (ANSI 92 standard syntax):

SELECT * 
FROM employee 
     RIGHT OUTER JOIN 
     department 
       ON employee.DepartmentID = department.DepartmentID

LastName	DepartmentID	DepartmentName	DepartmentID
Smith	34	Clerical	34
Jones	33	Engineering	33
Robinson	34	Clerical	34
Steinberg	33	Engineering	33
Rafferty	31	Sales	31
NULL	NULL	Marketing	35

Full outer join

A full outer join combines the results of both left and right outer joins. These joins will show records from both tables, and fill in NULLs for missing matches on either side.

Some database systems do not support this functionality, but it can be emulated through the use of left outer joins and unions.

Example full outer join (ANSI 92 standard syntax):

SELECT *  
FROM employee 
     FULL OUTER JOIN 
     department 
       ON employee.DepartmentID = department.DepartmentID

LastName	DepartmentID	DepartmentName	DepartmentID
Smith	34	Clerical	34
Jones	33	Engineering	33
Robinson	34	Clerical	34
Jasper	36	NULL	NULL
Steinberg	33	Engineering	33
Rafferty	31	Sales	31
NULL	NULL	Marketing	35

Implementation

The efficient implementation of joins has been the goal of much work in database systems, because joins are both extremely common and rather difficult to execute efficiently. The difficulty results from the fact that (inner) joins are both commutative and associative. In practice, this means that the user merely supplies the list of tables to be joined and the join conditions to be used, and the database system has the task of determining the most efficient way to perform the operation. Determining how to execute a query containing joins is done by the query optimizer. It has two basic freedoms:

1. join order: because joins are commutative, the order in which tables are joined does not change the final result set of the query. However, join order does have an enormous impact on the cost of the join operation, so choosing the right join order is very important.
2. join method: given two tables and a join condition, there are multiple algorithms to produce the result set of the join. Which algorithm is most efficient depends on the sizes of the input tables, the number of rows from each table that match the join condition, and the operations required by the rest of the query.

Many join algorithms treat their input tables differently. The input tables are referred to as the outer and inner tables, or left and right, respectively. In the case of nested loops, for example, the entire inner table will be scanned for each row of the outer table .

Query plans involving joins can be classified as:

• left-deep: the inner table of each join in the plan is a base table (rather than another join).
• right-deep: the outer table of each join in the plan is a base table.
• bushy: neither left-deep nor right-deep; both input tables of a join query may be joins themselves.

These names are derived from the appearance of the query if drawn as a tree, with the outer join relation on the left and the inner table on the right (as is the convention).

Join optimization

See Query optimizer

Join algorithms

There are four fundamental algorithms to perform a join operation.

Nested loops

This is the simplest join algorithm. For each tuple in the outer join relation, the entire inner join relation is scanned, and any tuples that match the join condition are added to the result set. Naturally, this algorithm performs poorly if either the inner or outer join relation is very large.

A refinement to this technique is called "block nested loops": for every block in the outer relation, the entire inner relation is scanned. For each match between the current inner tuple and one of the tuples in the current block of the outer relation, a tuple is added to the join result set. This variant means that more computation is done for each tuple of the inner relation, but far fewer scans of the inner relation are required.

Merge join

If both join relations are sorted by the join attribute, the join can be performed trivially:

1. For each tuple in the outer relation,
   1. Consider the current "group" of tuples from the inner relation; a group consists of a set of contiguous tuples in the inner relation with the same value in the join attribute.
   2. For each matching tuple in the current inner group, add a tuple to the join result. Once the inner group has been exhausted, both the inner and outer scans can be advanced to the next group.

This is one reason why many optimizers keep track of the sort order of query nodes — if one or both input relations to a merge join is already sorted on the join attribute, an additional sort is not required. Otherwise, the DBMS will need to perform the sort, usually using an external sort to avoid consuming too much memory.

See also Sort-Merge Join

Hash join

See Hash Join

Semi join

A semi join is an optimization technique for joins on distributed databases. The join predicates are applied in multiple phases, starting with the earliest possible. This can reduce the size of the intermediate results that must be exchanged with remote nodes, thus reducing inter node network traffic. It can be improved with a Bloom filter (hashing).

The technique as it relates to relational databases was first discussed in a 1981 Journal of the ACM article by Philip A. Bernstein and Dah-Ming W. Chiu titled Using semi-joins to solve relational queries.

See also

• SQL
• Select (SQL)
• Insert (SQL)

External links

• Various Join Algorithm Implementationscs:Join (SQL)

de:Relationale Algebra fr:Jointure ru:Алгоритм соединения (СУБД)