Detailed Oracle connection (connection left and right connections, fully connected ...)
1 Description:
Mentioned database must know the join operation, the contents of this part of the account for a pivotal position in the database query operations, today I join those who say the operation.
Coupling operation may be specified in the WHERE clause can be specified in the FROM clause, when the join condition specified in the FROM clause, the join operator SQL2 coupling type and is divided into two part coupling conditions.
Coupling type determines how to handle the join condition not matching tuple.
2 connecting operation:
Join condition determines the relationship between those two elements of the match.
| Join Type |
Explanation |
| INNER JOIN |
The coupling, the two coupling rows matching result table join |
| LEFT OUTER JOIN |
Left join: the result includes the left table (leftmost appear in the JOIN clause) all lines, not including non-matching rows in the right table. |
| RIGHT OUTER JOIN |
Right connection: The results include the right table (far right appear in the JOIN clause) in all rows, not including the line does not match the left table. |
| FULL OUTER JOIN |
Fully engaged: The results include all lines all join in, regardless of whether they match. |
| CROSS JOIN |
Cross coupling: the results of all possible combinations of the two coupling rows in the table. Cross-connect it returns Cartesian product of two tables. (Oracle does not support) |
| NATURAL JOIN |
Look for the column name and data type are equal fields in the two tables in the event of natural connection and automatically connects them together. |
Popular to say: when the left and right connections are at a table as a base table, all the contents of the table will display, and then add the contents of two tables match.
This figure references to: http: //blog.csdn.net/caolaosanahnu/article/details/8080350
For external connection, may be used in Oracle (+) to indicate, method used:
1, (+) operator can only occur in the WHERE clause, and can not be used simultaneously, and outer join syntax.
2, performed when the external connection (+) operator, if a plurality of conditions contained in the where clause, must be included in all conditions (+) operator.
3, (+) operator can only be applied to the column, and not be used in the expression.
4, (+) operator and can not be used in conjunction with or operator.
5, (+) operator can only be used to implement the outer connecting the left and right outer, external connection can not be used completely achieved.
Precautions natural connection:
(1) If the two do natural connection table has a plurality of fields with the same names and types to meet, the conditions as they will be connected.
(2) If two tables only natural identical field names, but not the same data type, then an error is returned.
Here I will explain a few connections, but also as a complement to the knowledge:
. 1, Self joins from the connection:
Since the use of the connection, a mirror with its own table as another table may be treated, after the join query data.
2, of Cartesian Products's Cartesian product
When two tables are not connected operation, the two tables of data query to get what it? It is the Cartesian product of two tables.
TABLE join operation over three
When the joint multi-table lookup is performed, you will find multi-table join operation will be carried out, and these connections operate in three modes:
Hash Join (hash join), Nested Loops, Sort Merge join, to understand these connections can help us to learn more about the join operation, the following were introduced one by one.
3.1 HASH JOIN (hash join)
Hash Join hash join mainly CBO (CBO: Cost-Based Optimization cost-based optimizer, the optimizer one Oracle, the subsequent description will be) larger than the data do when the connecting operation of the more common way, the optimizer will use two smaller tables that table (a data source or smaller table), using linkages (JOIN kEY) establishing in memory hash table, the column data stored in the hash list, and the large table scans, the same JOIN KEY to the table to hash matching, you can find out the value of the match, this time Note: If the HASH table is relatively large, not once stored into memory will be divided into no more partition (segment), temporary segment written to disk, At this time, the performance will be more than the cost of a written, may reduce efficiency.
This mainly applies to the smaller table (can be placed in memory), while the relevant cost performance is the sum of the two access performance.
Can USE_HASH (table_name1 table_name2) hint to force using a hash join.
3.2 SORT MERGE JOIN: sort-merge joins
Merge Join sort-merge join, the two tables are first sorted according to the associated keys (JOIN KEY), and extracting data from the respective sorting table, to another sorting table match.
Relatively speaking, merge join need to complete the sorting operation, it is more and more consumed by the performance, but if the source has been sorted, it will get better performance. Applies: inequivalent association (>, <,> =, <=, <>), HASH_JOIN_ENABLED = false and so on.
You may be used USE_MERGE (table_name1 table_name2) to force the use of sort-merge join.
3.3 NESTED LOOP: nested loop join
Nested Loops nested loop join, it is a cyclic mode of operation to read data from a table (table driven outer table), then access another table (look-up table inner table, indexed desired). Each row of the table link with a respective inner drive table records (JOIN). In the table is driven appearance, the appearance of each row in the table will be matched with the operation, so the query result set is best not to be too large. The small amount of data table as the driving table query (appearance), can be ordered using the default drive prompt CBO table.
For driving the table is small, nested loop join relatively good, while the inner table needs to access a valid index (index).
Use USE_NL (table_name1 table_name2) but forced CBO perform nested loop join.
Example 4:
Oracle's two tables: department information table: DEPTINFO, user table: USERINFO
4.1 INNER JOIN join operation:
Equivalent INNER JOIN connected as a connection, which is equivalent to the operation where the other is connected to.
4.2 LEFT JOIN connection operations:
USERINFO left connecting DEPTINFO, the result will USERINFO a base table, all content will be shown.
Oracle includes three kinds of connections: are:
4.3 RIGHT JOIN operation links
USERINFO DEPTINFO right connection, the result will DEPTINFO a base table, all of the contents will be displayed.
4.4 FULL OUTER JOIN connection operations:
USERINFO fully connected DEPTINFO, the result will USERINFO and content DEPTINFO are displayed.
4.5 NATURAL JION natural connection
4.6 using (+) to achieve connection operations:
1, no (+) is connected:
2, left (+) connection :( equivalent Right Join)
3, the right (+) :( connection corresponds Left Join)
Note: not all about plus (+).
4.7 Self Join a self-join and Cartesian product Cartesian Products
自连接 笛卡儿积
1 说明：
提到数据库必须要知道联接操作，这部分的内容在数据库查询操作中占到举足轻重的地位，今天我就来说说这些联结操作。
联接操作可以在WHERE中指定，也可以在FROM子句中指定，在FROM子句中指定联接条件时，SQL2将联接操作符分为联接类型和联接条件两部分。
联接类型决定了如何处理联接条件中不匹配的元组。
2 连接操作：
联接条件决定了两个关系中那些元素该匹配。
| 联接类型 |
说明 |
| INNER JOIN |
内联接，结果为两个联接表中的匹配行的联接 |
| LEFT OUTER JOIN |
左联接：结果包括左表（出现在JOIN子句最左边）中的所有行，不包括右表中的不匹配行。 |
| RIGHT OUTER JOIN |
右联接：结果包括右表（出现在JOIN子句最右边）中的所有行，不包括有左表中的不匹配的行。 |
| FULL OUTER JOIN |
完全联接：结果包括所有联接中的所有行，不论他们是否匹配。 |
| CROSS JOIN |
交叉联接：结果包括两个联接表中的所有可能的行组合。交叉连接返回的是两个表的笛卡儿积。（Oracle不支持） |
| NATURAL JOIN |
自然连接时在两张表中寻找那些数据类型 和列名都相等的字段，然后自动地将他们连接起来。 |
通俗的说：在左连接和右连接时都会以一张表作为基表，该表的内容会全部的显示，然后加上两张表匹配的内容。
此图引用至：http://blog.csdn.net/caolaosanahnu/article/details/8080350
对于外连接，在Oracle中可以使用(+)来表示，使用的方式：
1、(+)操作符只可以出现在WHERE子句中，并且不可以和outer join语法同时使用。
2、当使用(+)操作符执行外连接时，如果在where子句中包含多个条件，则必须在所有的条件中都包含(+)操作符。
3、(+)操作符只能适用于列，而不能用在表达式上。
4、(+)操作符不能与or和in操作符一起使用。
5、(+)操作符只能用于实现左外连接和右外连接，而不能用于实现完全外连接。
自然连接的注意事项：
（1） 如果做自然连接的两个表中有多个字段都满足有相同名称和类型，那么他们都会被作为连接的条件。
（2） 如果自然两个表中仅字段名称相同，但是数据类型不同时，此时会返回一个错误。
这里我再说明几种连接方式，也当作一种补充知识了：
1、 Self joins自连接：
使用自连接，可以将自己表的一个镜像作为另一张表来对待，进行连接后查询出数据。
2、 Cartesian Products 笛卡尔积
当两个表没有连接操作时，对这两个表进行查询得到的数据是什么样的呢？是这两个表的笛卡儿积。
3 多表连接操作
在进行多表联合查找时，会发现多表会进行连接操作的，而这些连接操作包括三种方式：
Hash Join（散列连接）、Nested Loops、Sort Merge join，对于这些连接方式的了解可以方便我们对于连接操作的进一步了解，下面进行一一介绍。
3.1 HASH JOIN（散列连接）
Hash Join散列连接主要是CBO（CBO: Cost-Based Optimization 基于代价的优化器， Oracle优化器之一，后续会进行介绍）做数据量比较大时进行连接操作比较常用的一种方式，优化器会选用两个表中较小那个表（小一点表或者数据源），利用连接键（JOIN KEY）在内存中建立散列表，将列数据存储到hash列表中，然后扫描大的那个表，同样对JOIN KEY到hash表进行匹配，找出可以匹配的值，此时注意：如果HASH表比较大，无法一次存储到内存时则会分成不多个partition(段)，写入磁盘的temporary segment，此时性能会多出一次写的代价，有可能会降低效率。
该内容主要适用于较小的表（可以放入内存），此时相关效能就是访问了两个的性能的成本总和。
可以用USE_HASH(table_name1 table_name2)提示来强制使用散列连接。
3.2 SORT MERGE JOIN：排序合并连接
Merge Join排序合并连接，是先将两个关联的表按照关联键（JOIN KEY）进行排序，然后从各自排序表中抽取数据，到另一个排序表中进行匹配。
相对来说，merge join需要完成排序操作，所以消耗的性能比较多，但是当源如果已经进行了排序，其将会取得较好的性能。适用于：不等价关联(>,<,>=,<=,<>)、HASH_JOIN_ENABLED=false等情况。
可以使用USE_MERGE(table_name1 table_name2)来强制使用排序合并连接.
3.3 NESTED LOOP：嵌套循环连接
Nested Loops嵌套循环连接，工作方式是循环从一张表中读取数据（驱动表 outer table）,然后访问另一张表（被查找表 inner table， 希望有索引）。驱动表中每一行与inner表中相应记录进行链接（JOIN）。内表被外表驱动，外表中的每一行都会与内表进行匹配操作，所以查询结果集最好不要太大。将数据量小的表作为查询的驱动表（外表），可以利用ordered 来提示CBO默认的驱动表。
对于驱动表较小的情况时，嵌套循环连接相对比较好，同时inner表需要有效的访问索引（index）。
使用USE_NL(table_name1 table_name2)可是强制CBO 执行嵌套循环连接。
4 举例:
Oracle的两张表：部门信息表：DEPTINFO，用户表：USERINFO
4.1 INNER JOIN连接操作：
INNER JOIN连接作为一种等值连接，其操作相当于where中的等至连接。
4.2 LEFT JOIN连接操作：
USERINFO 左连接DEPTINFO，结果将以USERINFO为基表，其内容将全部显示。
Oracle包含了3种连接方式：分别是：
4.3 RIGHT JOIN链接操作
USERINFO右连接DEPTINFO，结果将以DEPTINFO为基表，将其内容全部显示出来。
4.4 FULL OUTER JOIN连接操作：
USERINFO 全连接DEPTINFO，结果将USERINFO以及DEPTINFO的内容都显示出来。
4.5 NATURAL JION自然连接
4.6 利用(+)来实现连接操作:
1、无(+)连接：
2、左(+)连接：（相当于Right Join）
3、右(+)连接：（相当于Left Join）
注意：不可以左右都加(+)。
4.7 Self Join 自连接以及Cartesian Products笛卡儿积
自连接 笛卡儿积