在Hive0.11.0版本中新引进了一个新的特性，当用户将Hive查询结果输出到文件中时，用户可以指定列的分割符，而在之前的版本是不能指定列之间的分隔符。
在Hive0.11.0之前版本如下使用，无法指定分隔符，默认为\x01：
hive (hive)> insertoverwrite local directory '/home/hadoop/export_hive' select * from a;
Query ID =hadoop_20150627174342_64852f3a-56ed-48d5-a545-fc28f109be74
Total jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0since there's no reduce operator
Starting Job = job_1435392961740_0025,Tracking URL = http://gpmaster:8088/proxy/application_1435392961740_0025/
Kill Command =/home/hadoop/hadoop-2.6.0/bin/hadoop job -kill job_1435392961740_0025
Hadoop job information for Stage-1:number of mappers: 1; number of reducers: 0
2015-06-27 17:43:54,395 Stage-1 map =0%, reduce = 0%
2015-06-27 17:44:07,615 Stage-1 map =100%, reduce = 0%, Cumulative CPU 4.82sec
MapReduce Total cumulative CPU time: 4seconds 820 msec
Ended Job = job_1435392961740_0025
Copying data to local directory /home/hadoop/export_hive
Copying data to local directory/home/hadoop/export_hive
MapReduce Jobs Launched:
Stage-Stage-1: Map: 1 Cumulative CPU: 4.82 sec HDFS Read: 2416833 HDFS Write: 1188743SUCCESS
Total MapReduce CPU Time Spent: 4seconds 820 msec
OK
a.key a.value
Time taken: 26.475 seconds
(目录不存在时，会自动创建)
查看生成的文件内容：
[hadoop@gpmaster export_hive]$ head-n 10 /home/hadoop/export_hive/000000_0 | cat -A
2610^AaAAnz$
32196^AaAAoWnz$
78606^AaAAyXFz$
3804^AaAAz$
30102^AaABEWez$
21744^AaABukz$
39666^AaABz$
1632^AaABz$
82464^AaABz$
88320^AaACCaz$
我使用cat -A参数，将文件中每行的结尾$符号和分隔符^A(即是\x01)打印了出来。
接下来，我们使用Hive0.11.0版本新引进的新特性，指定输出结果列之间的分隔符：
hive (hive)> insertoverwrite local directory '/home/hadoop/export_hive' row format delimitedfields terminated by '*' select * from a;
Query ID =hadoop_20150627180045_fced1513-8f1b-44a8-8e88-3cd678552aa5
Total jobs = 1
Launching Job 1 out of 1
Number of reduce tasks is set to 0since there's no reduce operator
Starting Job = job_1435392961740_0028,Tracking URL = http://gpmaster:8088/proxy/application_1435392961740_0028/
Kill Command =/home/hadoop/hadoop-2.6.0/bin/hadoop job -kill job_1435392961740_0028
Hadoop job information for Stage-1:number of mappers: 1; number of reducers: 0
2015-06-27 18:00:57,354 Stage-1 map =0%, reduce = 0%
2015-06-27 18:01:10,567 Stage-1 map =100%, reduce = 0%, Cumulative CPU 4.68sec
MapReduce Total cumulative CPU time: 4seconds 680 msec
Ended Job = job_1435392961740_0028
Copying data to local directory/home/hadoop/export_hive
Copying data to local directory/home/hadoop/export_hive
MapReduce Jobs Launched:
Stage-Stage-1: Map: 1 Cumulative CPU: 4.68 sec HDFS Read: 2417042 HDFS Write: 1188743SUCCESS
Total MapReduce CPU Time Spent: 4seconds 680 msec
OK
a.key a.value
Time taken: 26.607 seconds
查看指定分隔符为*后，导出的数据如下：
[hadoop@gpmaster export_hive]$ head -n10 /home/hadoop/export_hive/000000_0
2610*aAAnz
32196*aAAoWnz
78606*aAAyXFz
3804*aAAz
30102*aABEWez
21744*aABukz
39666*aABz
1632*aABz
82464*aABz
88320*aACCaz
可以看到列的分隔符的确是我们指定的*号分隔符。
如果是复合类型，比如struct，map类型等也可以指定对应的分隔符：
以下我们做个实例操作实际操作以下：
(1) 创建复合类型的表
hive (hive)> create table userinfo(id int,name string,job_listarray
(2) 构造数据（使用默认分隔符构造）
1^A小明^AIT工程师^B教师^A10086^C正常^B10010^C不正常^A北京市^B130
2^A小花^A保姆^B护士^A10086^C正常^B10010^C正常^A南京市^B130
注释：
\001使用^A代替，\002使用^B，\003使用^C代替
造数据在使用vi编辑器里面，用ctrl+v然后再ct