HIVE中的order by操作

hive中常见的高级查询包括：group by、Order by、join、distribute by、sort by、cluster by、Union all。今天我们来看看order by操作，Order by表示按照某些字段排序，语法如下：

1. select col,col2...
2. from tableName
3. where condition
4. order by col1,col2 [asc|desc]

注意：

(1)：order by后面可以有多列进行排序，默认按字典排序。

(2)：order by为全局排序。

(3)：order by需要reduce操作，且只有一个reduce，无法配置(因为多个reduce无法完成全局排序)。

order by操作会受到如下属性的制约：

1. set hive.mapred.mode=nonstrict; (default value / 默认值)
2. set hive.mapred.mode=strict;

注：如果在strict模式下使用order by语句，那么必须要在语句中加上limit关键字，因为执行order by的时候只能启动单个reduce，如果排序的结果集过大，那么执行时间会非常漫长。

下面我们通过一个示例来深入体会order by的用法：

数据库有一个employees表，数据如下：

1. hive> select * from employees;
2. OK
3. lavimer 15000.0 ["li","lu","wang"]  {"k1":1.0,"k2":2.0,"k3":3.0}    {"street":"dingnan","city":"ganzhou","num":101} 2015-01-24  love
4. liao    18000.0 ["liu","li","huang"]    {"k4":2.0,"k5":3.0,"k6":6.0}    {"street":"dingnan","city":"ganzhou","num":102} 2015-01-24  love
5. zhang   19000.0 ["xiao","wen","tian"]   {"k7":7.0,"k8":8.0,"k8":8.0}    {"street":"dingnan","city":"ganzhou","num":103} 2015-01-24  love

现在我要按第二列(salary)降序排列：

1. hive> select * from employees order by salary desc;
2. //执行MapReduce的过程
3. Job 0: Map: 1  Reduce: 1   Cumulative CPU: 2.62 sec   HDFS Read: 415 HDFS Write: 245 SUCCESS
4. Total MapReduce CPU Time Spent: 2 seconds 620 msec
5. OK
6. zhang   19000.0 ["xiao","wen","tian"]   {"k7":7.0,"k8":8.0} {"street":"dingnan","city":"ganzhou","num":103} 2015-01-24  love
7. liao    18000.0 ["liu","li","huang"]    {"k4":2.0,"k5":3.0,"k6":6.0}    {"street":"dingnan","city":"ganzhou","num":102} 2015-01-24  love
8. lavimer 15000.0 ["li","lu","wang"]  {"k1":1.0,"k2":2.0,"k3":3.0}    {"street":"dingnan","city":"ganzhou","num":101} 2015-01-24  love
9. Time taken: 20.484 seconds
10. hive>

此时的hive.mapred.mode属性为：

1. hive> set hive.mapred.mode;
2. hive.mapred.mode=nonstrict
3. hive>

现在我们将它改为strict，然后再使用order by进行查询：

1. hive> set hive.mapred.mode=strict;
2. hive> select * from employees order by salary desc;
3. FAILED: Error in semantic analysis: 1:33 In strict mode, if ORDER BY is specified, LIMIT must also be specified. Error encountered near token 'salary'
4. hive>

注：在strict模式下查询必须加上limit关键字。

1. hive> select * from employees order by salary desc limit 3;
2. FAILED: Error in semantic analysis: No partition predicate found for Alias "employees" Table "employees"

注：另外还有一个要注意的是strict模式也会限制分区表的查询，解决方案是必须指定分区

先来看看分区：

1. hive> show partitions employees;
2. OK
3. date_time=2015-01-24/type=love
4. Time taken: 0.096 seconds

在strict模式先使用order by查询：

1. hive> select * from employees where partition(date_time='2015-01-24',type='love') order by salary desc limit 3;
2. FAILED: Parse Error: line 1:30 cannot recognize input near 'partition' '(' 'date_time' in expression specification
3. hive
4. > select * from employees where date_time='2015-01-24' and type='love' order by salary desc limit 3;
5. //执行MapReduce程序
6. Total MapReduce CPU Time Spent: 3 seconds 510 msec
7. OK
8. zhang   19000.0 ["xiao","wen","tian"]   {"k7":7.0,"k8":8.0} {"street":"dingnan","city":"ganzhou","num":103} 2015-01-24  love
9. liao    18000.0 ["liu","li","huang"]    {"k4":2.0,"k5":3.0,"k6":6.0}    {"street":"dingnan","city":"ganzhou","num":102} 2015-01-24  love
10. lavimer 15000.0 ["li","lu","wang"]  {"k1":1.0,"k2":2.0,"k3":3.0}    {"street":"dingnan","city":"ganzhou","num":101} 2015-01-24  love
11. Time taken: 19.861 seconds
12. hive>