大数据Hive Join连接查询

join_table:
    table_reference [INNER] JOIN table_factor [join_condition]
  | table_reference {LEFT|RIGHT|FULL} [OUTER] JOIN table_reference join_condition
  | table_reference LEFT SEMI JOIN table_reference join_condition
  | table_reference CROSS JOIN table_reference [join_condition] (as of Hive 0.10)
table_reference:
    table_factor
  | join_table
table_factor:
    tbl_name [alias]
  | table_subquery alias
  | ( table_references )
join_condition:
    ON expression
table_reference：是join查询中使用的表名，也可以是子查询别名（查询结果当成表参与join）。
table_factor：与table_reference相同,是联接查询中使用的表名,也可以是子查询别名。
join_condition：join查询关联的条件， 如果在两个以上的表上需要连接，则使用AND关键字。

SELECT *
FROM table1 t1, table2 t2, table3 t3
WHERE t1.id = t2.id AND t2.id = t3.id AND t1.zipcode = '02535';
从Hive 2.2.0开始，支持ON子句中的复杂表达式，支持不相等连接（请参阅HIVE-15211和HIVE-15251）。在此之前，Hive不支持不是相等条件的联接条件。
SELECT a.* FROM a JOIN b ON (a.id = b.id)
SELECT a.* FROM a JOIN b ON (a.id = b.id AND a.department = b.department)
SELECT a.* FROM a LEFT OUTER JOIN b ON (a.id <> b.id)

--table1: 员工表
CREATE TABLE employee(
   id int,
   name string,
   deg string,
   salary int,
   dept string
 ) row format delimited
fields terminated by ',';
--table2:员工住址信息表
CREATE TABLE employee_address (
    id int,
    hno string,
    street string,
    city string
) row format delimited
fields terminated by ',';
--table3:员工联系方式表
CREATE TABLE employee_connection (
    id int,
    phno string,
    email string
) row format delimited
fields terminated by ',';
--加载数据到表中
load data local inpath '/root/hivedata/employee.txt' into table employee;
load data local inpath '/root/hivedata/employee_address.txt' into table employee_address;
load data local inpath '/root/hivedata/employee_connection.txt' into table employee_connection;

--1、inner join
select e.id,e.name,e_a.city,e_a.street
from employee e inner join employee_address e_a
on e.id =e_a.id;
--等价于 inner join=join
select e.id,e.name,e_a.city,e_a.street
from employee e join employee_address e_a
on e.id =e_a.id;
--等价于 隐式连接表示法
select e.id,e.name,e_a.city,e_a.street
from employee e , employee_address e_a
where e.id =e_a.id;

--2、left join
select e.id,e.name,e_conn.phno,e_conn.email
from employee e left join employee_connection e_conn
on e.id =e_conn.id;
--等价于 left outer join
select e.id,e.name,e_conn.phno,e_conn.email
from employee e left outer join  employee_connection e_conn
on e.id =e_conn.id;

--3、right join
select e.id,e.name,e_conn.phno,e_conn.email
from employee e right join employee_connection e_conn
on e.id =e_conn.id;
--等价于 right outer join
select e.id,e.name,e_conn.phno,e_conn.email
from employee e right outer join employee_connection e_conn
on e.id =e_conn.id;

--4、full outer join
select e.id,e.name,e_a.city,e_a.street
from employee e full outer join employee_address e_a
on e.id =e_a.id;
--等价于
select e.id,e.name,e_a.city,e_a.street
from employee e full  join employee_address e_a
on e.id =e_a.id;

--5、left semi join
select *
from employee e left semi join employee_address e_addr
on e.id =e_addr.id;
--相当于 inner join 只不过效率高一些
select e.*
from employee e inner join employee_address e_addr
on e.id =e_addr.id;

--6、cross join
--下列A、B、C 执行结果相同，但是效率不一样：
--A:
select a.*,b.* from employee a,employee_address b where a.id=b.id;
--B:
select * from employee a cross join employee_address b on a.id=b.id;
select * from employee a cross join employee_address b where a.id=b.id;
--C:
select * from employee a inner join employee_address b on a.id=b.id;
--一般不建议使用方法A和B，因为如果有WHERE子句的话，往往会先进行笛卡尔积返回数据然后才根据WHERE条件从中选择。
--因此，如果两个表太大，将会非常非常慢，不建议使用。

SELECT a.* FROM a JOIN b ON (a.id = b.id)
SELECT a.* FROM a JOIN b ON (a.id = b.id AND a.department = b.department)
SELECT a.* FROM a LEFT OUTER JOIN b ON (a.id <> b.id)
b)  同一查询中可以连接2个以上的表
SELECT a.val, b.val, c.val FROM a JOIN b ON (a.key = b.key1) JOIN c ON (c.key = b.key2)
c)  如果每个表在联接子句中使用相同的列，则Hive将多个表上的联接转换为单个MR作业
SELECT a.val, b.val, c.val FROM a JOIN b ON (a.key = b.key1) JOIN c ON (c.key = b.key1)
--由于联接中仅涉及b的key1列，因此被转换为1个MR作业来执行
SELECT a.val, b.val, c.val FROM a JOIN b ON (a.key = b.key1) JOIN c ON (c.key = b.key2)
--会转换为两个MR作业，因为在第一个连接条件中使用了b中的key1列，而在第二个连接条件中使用了b中的key2列。第一个map / reduce作业将a与b联接在一起，然后将结果与c联接到第二个map / reduce作业中。
d)  join时的最后一个表会通过reducer流式传输，并在其中缓冲之前的其他表，因此，将大表放置在最后有助于减少reducer阶段缓存数据所需要的内存
SELECT a.val, b.val, c.val FROM a JOIN b ON (a.key = b.key1) JOIN c ON (c.key = b.key1)
--由于联接中仅涉及b的key1列，因此被转换为1个MR作业来执行，并且表a和b的键的特定值的值被缓冲在reducer的内存中。然后，对于从c中检索的每一行，将使用缓冲的行来计算联接。
SELECT a.val, b.val, c.val FROM a JOIN b ON (a.key = b.key1) JOIN c ON (c.key = b.key2)
--计算涉及两个MR作业。其中的第一个将a与b连接起来，并缓冲a的值，同时在reducer中流式传输b的值。
在第二个MR作业中，将缓冲第一个连接的结果，同时将c的值通过reducer流式传输。
e)  在join的时候，可以通过语法STREAMTABLE提示指定要流式传输的表。如果省略STREAMTABLE提示，则Hive将流式传输最右边的表。
SELECT /*+ STREAMTABLE(a) */ a.val, b.val, c.val FROM a JOIN b ON (a.key = b.key1) JOIN c ON (c.key = b.key1)
--a,b,c三个表都在一个MR作业中联接，并且表b和c的键的特定值的值被缓冲在reducer的内存中。然后，对于从a中检索到的每一行，将使用缓冲的行来计算联接。如果省略STREAMTABLE提示，则Hive将流式传输最右边的表。
f)  join在WHERE条件之前进行。
g)  如果除一个要连接的表之外的所有表都很小，则可以将其作为仅map作业执行
SELECT /*+ MAPJOIN(b) */ a.key, a.value FROM a JOIN b ON a.key = b.key
--不需要reducer。对于A的每个Mapper，B都会被完全读取。限制是不能执行FULL / RIGHT OUTER JOIN b。