简单的word-count操作：
[root@master test-map]# head -10 The_Man_of_Property.txt #先看看数据
Preface
“The Forsyte Saga” was the title originally destined for that part of it which is called “The Man of Property”; and to adopt it for the collected chronicles of the Forsyte family has indulged the Forsytean tenacity that is in all of us. The word Saga might be objected to on the ground that it connotes the heroic and that there is little heroism in these pages. But it is used with a suitable irony; and, after all, this long tale, though it may deal with folk in frock coats, furbelows, and a gilt-edged period, is not devoid of the essential heat of conflict. Discounting for the gigantic stature and blood-thirstiness of old days, as they have come down to us in fairy-tale and legend, the folk of the old Sagas were Forsytes, assuredly, in their possessive instincts, and as little proof against the inroads of beauty and passion as Swithin, Soames, or even Young Jolyon. And if hero
hive> create table article2(sentence string) row format delimited fields terminated by ' '; --直接用行号为分隔符导进来。
hive> load data inpath '/The_Man_of_Property.txt' overwrite into table article2; --这个不加local的情况下是在本地导入的，即从hsfs中的/下导入，但是导入进去之后The_Man_of_Property.txt文件会消失。
hive> select explode(split(sentence,' ')) from article2 limit 10; --这个是用explode函数将sentence列，用空格分开。
OK
Preface
“The
Forsyte
hive> select word ,count(*) from (select explode(split(sentence,' ')) as word from article2 ) t group by word limit 10;
Query ID = root_20180504132418_41a4e570-f3d5-4dd8-8a86-3c93256c4063
Total jobs = 1
Launching Job 1 out of 1
Number of reduce tasks not specified. Estimated from input data size: 1
In order to change the average load for a reducer (in bytes):
set hive.exec.reducers.bytes.per.reducer=<number>
In order to limit the maximum number of reducers:
set hive.exec.reducers.max=<number>
In order to set a constant number of reducers:
set mapreduce.job.reduces=<number>
Starting Job = job_1525332220382_0102, Tracking URL = http://master:8088/proxy/application_1525332220382_0102/
Kill Command = /usr/local/src/hadoop-2.6.1/bin/hadoop job -kill job_1525332220382_0102
Hadoop job information for Stage-1: number of mappers: 1; number of reducers: 1
2018-05-04 13:24:24,802 Stage-1 map = 0%, reduce = 0%
2018-05-04 13:24:32,114 Stage-1 map = 100%, reduce = 0%, Cumulative CPU 3.91 sec
2018-05-04 13:24:40,501 Stage-1 map = 100%, reduce = 100%, Cumulative CPU 6.59 sec
MapReduce Total cumulative CPU time: 6 seconds 590 msec
Ended Job = job_1525332220382_0102
MapReduce Jobs Launched:
Stage-Stage-1: Map: 1 Reduce: 1 Cumulative CPU: 6.59 sec HDFS Read: 640084 HDFS Write: 116 SUCCESS
Total MapReduce CPU Time Spent: 6 seconds 590 msec
OK
35
(Baynes 1
(Dartie 1
(Dartie’s 1
(Down-by-the-starn) 2
(Down-by-the-starn), 1
(He 1
(I 1
(James) 1
(L500) 1
Time taken: 23.514 seconds, Fetched: 10 row(s)
分桶的操作：
hive> set hive.enforce.bucketing = true; --不设置的话可能不生效。
hive> create table user_bucket(id int) clustered by (id) into 4 buckets; --人为的分成的4个reduce，即四个桶,使用id分桶。
hive> insert overwrite table user_bucket select cast(user_id as int) from badou.udata;
采样：
• 查看sampling数据：• 查看sampling数据：
– hive> select * from student tablesample(bucket 1 out of 32 on id); 2bucket=1y 32bucket=16y
– tablesample是抽样语句，语法：TABLESAMPLE(BUCKET x OUT OF y)
– y必须是table总bucket数的倍数或者因子。hive根据y的大小，决定抽样的比例。例如，table总共分了64份，当y=32
时，抽取(64/32=)2个bucket的数据，当y=128时，抽取(64/128=)1/2个bucket的数据。x表示从哪个bucket开始抽
取。例如，table总bucket数为32，tablesample(bucket 1 out of 16)，表示总共抽取（32/16=）2个bucket的数据，
分3第三个bucket，和第（3+16)19个bucket的数据。这个3是随机的，只要间隔是16就好
采样的例子：
之后我们来查看数据量:
采样的数据量
总数据量：
之后对比一下数量：
结果差不多一致。
创建内外部表：
然后将数据插入表中
查看hdfs上的路径，
之后删除内部表：
在HIve中查看表
再去hdfs上查看路径，发现没有这个路径了。
之后删除外部表：
表的内容消失了
之后查看hdfs上的路径：
发现外部表的hdfs上的数据并没消失，之后可以导入进去就可以重新，创建了。所以外部表比较安全。
分区表：
内部表：
hive>
create table partition_test(
order_id string,
user_id string,
eval_set string,
order_number string,
order_hourofday string,
order_since_last string
)partitioned by(order_week string)
row format delimited fields terminated by ' ';
在hdfs上查找一下：
hive>insert overwrite table partition_test partition (order_week='3')
select order_id,user_id,eval_set,order_number,order_since_last,order_since_last from orders where order_week='3' ;
将order_week=1，2，3分别作为一个分区：然后查看这个表；（这里只截取一部分）
然后器hdfs上查看路径，分成了三个区；
在大量数据的场景中，即（表中的数据以及每个分区的个数都非常大的话），一个高度的安全建议的就是将HIve设置为“strict（严格模式）”,这样如果对分区表进行查询而WHERE子句没有加分区过滤的话，将会禁止提交这个任务。
hive> set hive.mapred.mode=strict; --设置为严格模式；
hive> set hive.mapred.mode=nostrict --设置为非严格模式；
可以使用如下命令查看所有的分区表；
查看指定的分区键的分区；
使用如下命令查看分区表的信息：
hive>alter table partition_test add partition (order_week=0) location '/usr/hive2/warehouse/badou.db/partition_test/order_week=0'; --添加分区的操作。
外部表：
和内部表差不多，但外部表的优势就是删除表不会删除hdfs上的数据，而且可以共享数据。