Hadoop 架构初探

对流行Hadoop做了一些最基本的了解,暂时没太大感觉,恩先记点笔记吧. = =

Hadoop 基本命令及环境安装

一、下载虚拟机镜像

目前比较流行的有以下三个：

(CHD) http://www.cloudera.com
(HDP)  http://hortonworks.com/
(MapR) http://www.mapr.com

本文使用HDP的沙盘
下载地址 http://hortonworks.com/products/hortonworks-sandbox/#install
我使用的是 Hyper-V 的镜像 , 配置可以查看下载地址旁边的文档

二、使用HDP沙盘

1. 显示指定路径的文件和目录
   注意这里显示的hdfs的文件目录不是实际机器中的文件目录

   hadoop fs -ls /

2. 建立一个目录并下载数据

   mkdir /home/bihell
   wget  http://www.grouplens.org/system/files/ml-100k.zip
   unzip ml-100k.zip

3. 在Hadoop中建立目录

   hadoop fs -mkdir /bihell/
   hadoop fs -mkdir /bihell/movies
   hadoop fs -mkdir /bihell/userinfo

4. Hadoop文件操作
   
   hadoop支持两个文件系统命令

   fs put 命令可以把文件传送到hadoop的文件系统,而fs get 命令可以从hadoop中获取文件

   hadoop fs -put u.item /bihell/movies
   hadoop fs -put u.info /bihell/userinfo

   另外还有一个拷贝命令  fs –cp

   hadoop fs -cp /bihell/movies/u.item /bihell

   删除命令 fs -rm

   hadoop fs -rm  /bihell/u.item

   拷贝多个文件

   hadoop fs -mkdir /bihell/test
   hadoop fs -cp /bihell/movies/u.item /bihell/userinfo/u.info /bihell/test

   递归删除文件

   hadoop fs -rm -r -skipTrash /bihell/test

   显示文件内容

   hadoop fs -cat /bihell/movies/* |less

三、 使用hue ui 的文件浏览器操作文件

根据沙盘的提示访问 http://192.168.56.101:8000/filebrowser/#/  我们可以看到刚才建立的目录。 （还是UI方便点啊）

使用Hive并且将数据导入仓库

一、先看一下Demo里面的Hive目录

hadoop fs -ls /apps/hive/warehouse

Found 3 items
drwxrwxrwx   - hive hdfs          0 2015-08-20 09:05 /apps/hive/warehouse/sample_07
drwxrwxrwx   - hive hdfs          0 2015-08-20 09:05 /apps/hive/warehouse/sample_08
drwxrwxrwx   - hive hdfs          0 2015-08-20 08:58 /apps/hive/warehouse/xademo.db

hadoop fs -ls /apps/hive/warehouse/sample_07

Found 1 items
-rwxr-xr-x   1 hue hue      46055 2015-08-20 08:46 /apps/hive/warehouse/sample_07/sample_07

查看文件内容

hadoop fs -cat /apps/hive/warehouse/sample_07/sample_07 | less

二、使用hive命令

进入hive数据库

hive

显示hive中的数据库

show databases;

显示表格

show tables;
show tables '*08*';

清空屏幕

!clear;

进一步查看表格结构

describe sample_07;
describe extended sample_07 ;

创建数据库

create database bihell;

使用hadoop fs命令查看下hive 目录,我们刚才创建的数据库文件应该在里面了

!hadoop fs -ls /apps/hive/warehouse/;

结果如下:

Found 4 items
drwxrwxrwx   - root hdfs          0 2015-09-12 08:57 /apps/hive/warehouse/bihell.db
drwxrwxrwx   - hive hdfs          0 2015-08-20 09:05 /apps/hive/warehouse/sample_07
drwxrwxrwx   - hive hdfs          0 2015-08-20 09:05 /apps/hive/warehouse/sample_08
drwxrwxrwx   - hive hdfs          0 2015-08-20 08:58 /apps/hive/warehouse/xademo.db

三、使用建立的数据库

一直用命令行比较吃力,我们也可用ui界面

在我们新建的bihell数据库中建立表格

CREATE TABLE movies (
     movie_id INT,
     movie_title STRING,
     release_date STRING,
     video_release_date STRING,
     imdb_url STRING,
     unknown INT,
     action INT,
     adventure INT,
     animation INT,
     children INT,
     comedy INT,
     crime INT,
     documentary INT,
     drama INT,
     fantasy INT,
     film_noir INT,
     horror INT,
     musical INT,
     mystery INT,
     romance INT,
     sci_fi INT,
     thriller INT,
     war INT,
     Western INT
)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY '|'
STORED AS TEXTFILE;

创建完毕以后点击Tables可以看到我们刚才创建的表格

在SSH执行文件命令,我们可以看到bihell.db下面多了一个目录

hadoop fs -ls /apps/hive/warehouse/bihell.db

Found 1 items
drwxrwxrwx   - hive hdfs          0 2015-09-12 09:09 /apps/hive/warehouse/bihell.db/movies

四、进入hive ,我们导入一些数据进去

导入数据

lOAD DATA INPATH '/bihell/userinfo' INTO TABLE movies;

清空数据

truncate table movies;

导入并覆盖原有数据

load data inpath '/bihell/movies' overwrite into table movies;

四、建立External表与RCFile 表

前面我们建立表以后导入数据到表中, 目录中的文件会被删除,现在我们直接建立表并指向我们所在的文件目录,建立外部表.

复原文件

!hadoop fs -put /home/bihell/ml-100k/u.user /bihell/userinfo;

建立另外一个表格,注意有指定路径

CREATE EXTERNAL TABLE users (
user_id INT,
age INT,
gender STRING,
occupation STRING,
zip_code STRING
)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY '|'
STORED AS TEXTFILE
LOCATION '/bihell/userinfo';

查看users的schema

describe formatted users;

查询表

SELECT * FROM users limit 100;

创建 RCFile 表格

CREATE TABLE occupation_count
STORED AS RCFile
AS SELECT COUNT(*), occupation FROM users GROUP BY occupation;

引用另外一个表创建一个空表

CREATE TABLE occupation2 LIKE occupation_count;

Hive 查询语言

我们之前已经用了部分hive查询,现在深入一下

一、复杂类型

Arrays – ARRAY<data_type>

Maps  -- MAP<primitive,data_type>

Struct  -- STRUCT<col_name:data_type[COMMENT col_comment],…>

Union Type – UNIONYTPE<data_type,data_type,…>

create table movies (

movie_name string,

participants ARRAY <string>,

release_dates MAP <string,timestamp>,

studio_addr STRUCT  <state:string,city:string,zip:string,streetnbr:int,streetname:string,unit:string>,

complex_participants MAP<string,STRUCT<address:string,attributes MAP<string,string>>>

misc UNIONTYPE <int,string,ARRAY<double>>

);

查询方式

select movie_name,
    participants[0],
    release_dates[“USA”],
    studio_addr.zip,
    complex_participants[“Leonardo DiCaprio”].attributes[“fav_color”],
    misc
from movies;

二、Partitioned Tables

这个章节主要讲述加载与管理Hive中的数据

前面我们使用了CREATE TABLE 以及 CREATE EXTERNAL TABLE 本文我们要看下Table Partitions

创建分区表:

CREATE TABLE page_views( eventTime STRING, userid STRING)
PARTITIONED BY (dt STRING, applicationtype STRING)
STORED AS TEXTFILE;

数据库文件默认地址 :

/apps/hive/warehouse/page_views

当你每次导入数据的时候都会为你建立partition ,比如

LOAD DATA INPATH ‘/mydata/android'/Aug_10_2013/pageviews/’
INTO TABLE page_views
PARTITION (dt = ‘2013-08-10’, applicationtype = ‘android’);

生成分区如下:

/apps/hive/warehouse/page_views/dt=2013-08-10/application=android

当然我们也可以覆盖导入

LOAD DATA INPATH ‘/mydata/android'/Aug_10_2013/pageviews/’
OVERWRITE INTO TABLE page_views
PARTITION (dt = ‘2013-08-10’, applicationtype = ‘android’);

创建语句中dt和applicationtype 是virtual partition columns. 如果你describe table,会发现所有字段显示和正常表一样

eventTime STRING

userid STRING

page STRING

dt STRING

applicationtype STRING

可以直接用于查询

select dt as eventDate,page,count(*) as pviewCount From page_views
where applicationtype = ‘iPhone’;

三、External Partitioned Tables

相比分区表,只是多了一个EXTERNAL ,我们注意到这里没有指定location ,添加文件的时候才需要指定

CREATE  EXTERNAL TABLE page_views( eventTime STRING, userid STRING)
PARTITIONED BY (dt STRING, applicationtype STRING)
STORED AS TEXTFILE;

添加文件

ALTER TABLE page_views ADD PARTITION ( dt = ‘2013-09-09’, applicationtype = ‘Windows Phone 8’)
LOCATION ‘/somewhere/on/hdfs/data/2013-09-09/wp8’;

ALTER TABLE page_view ADD PARTITION (dt=’2013-09-09’,applicationtype=’iPhone’)
LOCATION ‘hdfs://NameNode/somewhere/on/hdfs/data/iphone/current’;

ALTER TABLE page_views ADD IF NOT EXSTS
PARTITION (dt=’2013-09-09’,applicationtype=’iPhone’) LOCATION ‘/somewhere/on/hdfs/data/iphone/current’;
PARTITION (dt=’2013-09-08’,applicationtype=’iPhone’) LOCATION ‘/somewhere/on/hdfs/data/prev1/iphone;
PARTITION (dt=’2013-09-07’,applicationtype=’iPhone’) LOCATION ‘/somewhere/on/hdfs/data/iphone/prev2;

四、实际操作

EXTERNAL PARTITION TABLE

--建立目录
hadoop fs -mkdir /bihell/logs/pv_ext/somedatafor_7_11 /bihell/logs/pv_ext/2013/08/11/log/data

--建立EXTERNAL TABLE
CREATE EXTERNAL TABLE page_views_ext (logtime STRING, userid INT, ip STRING, page STRING, ref STRING, os STRING, os_ver STRING, agent STRING)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY '\t'
LOCATION '/bihell/logs/pv_ext/';

--查看表格详细信息
DESCRIBE FORMATTED page_views_ext;

--查看执行计划
EXPLAIN SELECT * FROM page_views_ext WHERE userid = 13;

--删除表
DROP TABLE page_views_ext;

--创建EXTERNAL Partition Table
CREATE EXTERNAL TABLE page_views_ext (logtime STRING, userid INT, ip STRING, page STRING, ref STRING, os STRING, os_ver STRING, agent STRING)
PARTITIONED BY (y STRING, m STRING, d STRING)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY '\t'
LOCATION '/bihell/logs/pv_ext/';

--将日志传送至Hadoop目录
!hadoop fs -put /media/sf_VM_Share/LogFiles/log_2013711_155354.log /bihell/logs/pv_ext/somedatafor_7_11

--因为是partition table 所以此时查询该表是没有任何内容的
SELECT * FROM page_views_ext;

--添加文件
ALTER TABLE page_views_ext ADD PARTITION (y='2013', m='07', d='11')
LOCATION '/bihell/logs/pv_ext/somedatafor_7_11';

--再次查询
SELECT * FROM page_views_ext LIMIT 100;

--describe table
DESCRIBE FORMATTED page_views_ext;

--再次查看执行计划
我们发现predicate还是13, 并没有加上 m,d
EXPLAIN SELECT * FROM page_views_ext WHERE userid=13 AND m='07'AND d='11' LIMIT 100;

--再添加一个文件
!hadoop fs -put /media/sf_VM_Share/LogFiles/log_2013811_16136.log /bihell/logs/pv_ext/2013/08/11/log/data
ALTER TABLE page_views_ext ADD PARTITION (y='2013', m='08', d='11')
LOCATION '/bihell/logs/pv_ext/2013/08/11/log/data';

--查询
SELECT COUNT(*) as RecordCount, m FROM page_views_ext WHERE d='11' GROUP BY m;

--另一种方式添加数据
!hadoop fs -put /media/sf_VM_Share/LogFiles/log_2013720_162256.log /bihell/logs/pv_ext/y=2013/m=07/d=20/data.log
SELECT * FROM page_views_ext WHERE m='07' AND d='20' LIMIT 100;
MSCK REPAIR TABLE page_views_ext;
SELECT * FROM page_views_ext WHERE m='07' AND d='20' LIMIT 100;

PARTITION TABLE

CREATE TABLE page_views (logtime STRING, userid INT, ip STRING, page STRING, ref STRING, os STRING, os_ver STRING, agent STRING)
PARTITIONED BY (y STRING, m STRING, d STRING)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY '\t';

LOAD DATA LOCAL INPATH '/media/sf_VM_Share/LogFiles/log_2013805_16210.log'
OVERWRITE INTO TABLE page_views PARTITION (y='2013', m='08', d='05');

!hadoop fs -ls /apps/hive/warehouse/bihell.db/page_views/;

批量插入及动态分区表插入

Multiple Inserts

--Syntax

FROM form_statement

INSERT OVERWRITE TABLE table1 [PARTITION(partcol1=val1,partcol2=val2)] select_statement1

INSERT INTO TABLE table2 [PARTITION(partcol1=val1,partcol2=val2)[IF NOT EXISTS]] select_statements2

INSERT OVERWRITE DIRECTORY ‘path’ select_statement3;

-- 提取操作
FROM movies
INSERT OVERWRITE TABLE horror_movies SELECT * WHERE horror = 1 AND release_date=’8/23/2013’
INSERT INTO action_movies SELECT * WHERE action = 1 AND release_date = ‘8/23/2013’;

FROM (SELECT * FROM movies WHERE release_date =’8/23/2013’) src
INSERT OVERWRITE TABLE horror_movies SELECT * WHERE horror =1
INSERT INTO action_movies SELECT * WHERE action = 1;

Dynamic Partition Inserts

CREATE TABLE views_stg (eventTime STRING, userid STRING)
PARTITIONED BY(dt STRING,applicationtype STRING,page STRING);

FROM page_views src
INSERT OVERWRITE TABLE views_stg PARTITION (dt=’2013-09-13’,applicationtype=’Web’,page=’Home’)
    SELECT src.eventTime,src.userid WHERE dt=’2013-09-13’ AND applicationtype=’Web’,page=’Home’
INSERT OVERWRITE TABLE views_stg PARTITION (dt=’2013-09-14,applicationtype=’Web’,page=’Cart’)
    SELECT src.eventTime,src.userid WHERE dt=’2013-09-14’ AND applicationtype=’Web’,page=’Cart’
INSERT OVERWRITE TABLE views_stg PARTITION (dt=’2013-09-15’,applicationtype=’Web’,page=’Checkout’)
    SELECT src.eventTime,src.userid WHERE dt=’2013-09-15’ AND applicationtype=’Web’,page=’Checkout’

FROM page_views src
INSERT OVERWRITE TABLE views_stg PARTITION (applicationtype=’Web’,dt,page)
SELECT src.eventTime,src.userid,src.dt,src.page WHERE applicationtype=’Web’

实例

!hadoop fs -mkdir /bihell/logs/multi_insert;

!hadoop fs -put /media/sf_VM_Share/LogFiles/log_2012613_161117.log /media/sf_VM_Share/LogFiles/log_2013803_15590.log /bihell/logs/multi_insert

-- 创建EXTERNAL TABLE
CREATE EXTERNAL TABLE staging (logtime STRING, userid INT, ip STRING, page STRING, ref STRING, os STRING, os_ver STRING, agent STRING)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY '\t'
LOCATION '/bihell/logs/multi_insert';

--批量插入 PARTITION
INSERT INTO TABLE page_views PARTITION (y, m, d)
SELECT logtime, userid, ip, page, ref, os, os_ver, agent, substr(logtime, 7, 4), substr(logtime, 1, 2), substr(logtime, 4, 2)
FROM staging;

SET hive.exec.dynamic.partition.mode=nonstrict;

INSERT INTO TABLE page_views PARTITION (y, m, d)
SELECT logtime, userid, ip, page, ref, os, os_ver, agent, substr(logtime, 7, 4), substr(logtime, 1, 2), substr(logtime, 4, 2)
FROM staging;

SELECT * FROM page_views WHERE y='2012' LIMIT 100;

select regexp_replace(logtime, '/', '-') from staging;
select substr(logtime, 7, 4), substr(logtime, 1, 2), substr(logtime, 4, 2) from staging;