hive的元数据要保存到mysql里，hdfs上就是文件，没有定义各种数据格式
将文件格式化读取，必须要有区分字符
hive依赖的关系数据库mysql，不用维护！

建库

hive> create database test;
hive> drop database test;

默认在default数据库

建表

LanguageManual DDL – Create/Drop/Truncate Table

hive> drop table psn;
hive> create table psn(
id int,
name string,
likes array<string>,
address map<string, string>
)
ROW FORMAT DELIMITED 
FIELDS TERMINATED BY ','
COLLECTION ITEMS TERMINATED BY '-'
MAP KEYS TERMINATED BY ':'
LINES TERMINATED BY '\n'; 

属性通过用什么隔开？(集合、map……)
行通过什么隔开？默认是'\n',可不写

外部表和内部表的区别

EXTERNAL_TABLE | MANAGED_TABLE
创建时需要指定目录
    location '/usr/';
删除时内部表将`表结构`和`数据`全部删除；外部表只删除`表结构`，不删除`数据`
    表结构：hive上的表
    数据：HDFS上数据文件
外部表、外部表元数据都保存在mysql中

查看表结构

hive> desc formatted psn;
OK
# col_name              data_type               comment             
         
id                      int                                         
name                    string                                      
likes                   array<string>                               
address                 map<string,string>                          
         
# Detailed Table Information         
Database:               default                  
Owner:                  root                     
CreateTime:             Sat Nov 24 08:43:34 CST 2018     
LastAccessTime:         UNKNOWN                  
Protect Mode:           None                     
Retention:              0                        
Location:               hdfs://mycluster/user/hive/warehouse/psn     
Table Type:             MANAGED_TABLE            
Table Parameters:        
    COLUMN_STATS_ACCURATE   true                
    numFiles                1                   
    totalSize               498                 
    transient_lastDdlTime   1543020308          
         
# Storage Information        
SerDe Library:          org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe   
InputFormat:            org.apache.hadoop.mapred.TextInputFormat     
OutputFormat:           org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat   
Compressed:             No                       
Num Buckets:            -1                       
Bucket Columns:         []                       
Sort Columns:           []                       
Storage Desc Params:         
    colelction.delim        -                   
    field.delim             ,                   
    line.delim              \n                  
    mapkey.delim            :                   
    serialization.format    ,                   
Time taken: 0.251 seconds, Fetched: 36 row(s)

上传并加载文件

LanguageManual DML – Loading files into tables
数据要对应，否则不能成功！

hive> LOAD DATA LOCAL INPATH '/root/psn.data' INTO TABLE psn;

Hive上查询

hive> select * from psn;
OK
1 小明1 ["lol","book","movie"] {"shenzhen":"luohu","shanghai":"pudong"}
2 小明2 ["lol","book","movie"] {"shenzhen":"luohu","shanghai":"pudong"}
3 小明3 ["lol","book","movie"] {"shenzhen":"luohu","shanghai":"pudong"}
4 小明4 ["lol","movie"] {"shenzhen":"luohu","shanghai":"pudong"}
5 小明5 ["lol","book","movie"] {"shenzhen":"luohu","shanghai":"pudong"}
6 小明6 ["lol","book","movie"] {"shenzhen":"luohu","shanghai":"pudong"}
7 小明7 ["lol","book","game"] {"shenzhen":"luohu","shanghai":"pudong"}
8 小明8 ["lol","book","movie"] {"shenzhen":"luohu","shanghai":"pudong"}
9 小明9 ["lol","book","movie"] {"shenzhen":"luohu","shanghai":"pudong"}

HDFS上查询

[root@node004 ~]# hdfs dfs -cat /user/hive/warehouse/psn/*
1,小明1,lol-book-movie,shenzhen:luohu-shanghai:pudong
2,小明2,lol-book-movie,shenzhen:luohu-shanghai:pudong
3,小明3,lol-book-movie,shenzhen:luohu-shanghai:pudong
4,小明4,lol-movie,shenzhen:luohu-shanghai:pudong
5,小明5,lol-book-movie,shenzhen:luohu-shanghai:pudong
6,小明6,lol-book-movie,shenzhen:luohu-shanghai:pudong
7,小明7,lol-book-game,shenzhen:luohu-shanghai:pudong
8,小明8,lol-book-movie,shenzhen:luohu-shanghai:pudong
9,小明9,lol-book-movie,shenzhen:luohu-shanghai:pudong

参考资料

Hadoop集群上搭建Hive – 简书