271 lines
8.5 KiB
Markdown
271 lines
8.5 KiB
Markdown
## Hive简介
|
||
|
||
[Hive](https://hive.apache.org/) 是 Facebook 开源的一款基于 Hadoop 的数据仓库工具，目前由 Apache 软件基金会维护，它是应用最广泛的大数据处理解决方案，它能将 SQL 查询转变为 MapReduce（Google提出的一个软件架构，用于大规模数据集的并行运算）任务，对 SQL 提供了完美的支持，能够非常方便的实现大数据统计。
|
||
|
||
<img src="res/sql_to_mr.png" style="zoom:50%;">
|
||
|
||
<img src="res/HADOOP-ECOSYSTEM-Edureka.png">
|
||
|
||
> **说明**：可以通过<https://www.edureka.co/blog/hadoop-ecosystem>来了解 Hadoop 生态圈。
|
||
|
||
如果要简单的介绍 Hive，那么以下两点是其核心：
|
||
|
||
1. 把 HDFS 中结构化的数据映射成表。
|
||
2. 通过把 HQL 进行解析和转换，最终生成一系列基于 Hadoop 的 MapReduce 任务或 Spark 任务，通过执行这些任务完成对数据的处理。也就是说，即便不学习 Java、Scala 这样的编程语言，一样可以实现对数据的处理。
|
||
|
||
Hive的应用场景。
|
||
|
||
<img src="res/what_hive_can_do.png" style="zoom:50%;">
|
||
|
||
<img src="res/what_hive_can_not_do.png" style="zoom:35%;">
|
||
|
||
Hive和传统关系型数据库的对比如下图和下表所示。
|
||
|
||
<img src="res/hive_vs_rdbms.png" style="zoom:50%;">
|
||
|
||
| | Hive | RDBMS |
|
||
| -------- | ----------------- | ------------ |
|
||
| 查询语言 | HQL | SQL |
|
||
| 存储数据 | HDFS | 本地文件系统 |
|
||
| 执行方式 | MapReduce / Spark | Executor |
|
||
| 执行延迟 | 高 | 低 |
|
||
| 数据规模 | 大 | 小 |
|
||
|
||
### 准备工作
|
||
|
||
1. 搭建如下图所示的大数据平台。
|
||
|
||
<img src="res/20220210080638.png" style="zoom:60%;">
|
||
|
||
2. 通过Client节点（跳板机）访问大数据平台。
|
||
|
||
<img src="res/20220210080655.png" style="zoom:50%;">
|
||
|
||
3. 创建文件Hadoop的文件系统。
|
||
|
||
```Shell
|
||
hdfs dfs -mkdir /user/root
|
||
```
|
||
|
||
4. 将准备好的数据文件拷贝到Hadoop文件系统中。
|
||
|
||
```Shell
|
||
hdfs dfs -put /home/ubuntu/data/* /user/root
|
||
```
|
||
|
||
5. 进入 hive 命令行。
|
||
|
||
```Shell
|
||
hive
|
||
```
|
||
|
||
### 建库建表
|
||
|
||
1. 创建。
|
||
|
||
```SQL
|
||
create database eshop;
|
||
```
|
||
|
||
2. 删除。
|
||
|
||
```SQL
|
||
drop database eshop cascade;
|
||
```
|
||
|
||
3. 切换。
|
||
|
||
```SQL
|
||
use eshop;
|
||
```
|
||
|
||
#### 数据类型
|
||
|
||
Hive的数据类型如下所示。
|
||
|
||
<img src="res/hive_data_types.png" style="zoom:50%;">
|
||
|
||
基本数据类型：
|
||
|
||
| 数据类型 | 占用空间 | 支持版本 |
|
||
| --------- | -------- | -------- |
|
||
| tinyint | 1-Byte | |
|
||
| smallint | 2-Byte | |
|
||
| int | 4-Byte | |
|
||
| bigint | 8-Byte | |
|
||
| boolean | | |
|
||
| float | 4-Byte | |
|
||
| double | 8-Byte | |
|
||
| string | | |
|
||
| binary | | 0.8版本 |
|
||
| timestamp | | 0.8版本 |
|
||
| decimal | | 0.11版本 |
|
||
| char | | 0.13版本 |
|
||
| varchar | | 0.12版本 |
|
||
| date | | 0.12版本 |
|
||
|
||
复合数据类型：
|
||
|
||
| 数据类型 | 描述 | 例子 |
|
||
| -------- | ------------------------ | --------------------------------------------- |
|
||
| struct | 和C语言中的结构体类似 | `struct<first_name:string, last_name:string>` |
|
||
| map | 由键值对构成的元素的集合 | `map<string,int>` |
|
||
| array | 具有相同类型的变量的容器 | `array<string>` |
|
||
|
||
4. 创建内部表。
|
||
|
||
```SQL
|
||
create table if not exists dim_user_info
|
||
(
|
||
user_id string,
|
||
user_name string,
|
||
sex string,
|
||
age int,
|
||
city string,
|
||
firstactivetime string,
|
||
level int,
|
||
extra1 string,
|
||
extra2 map<string,string>
|
||
)
|
||
row format delimited fields terminated by '\t'
|
||
collection items terminated by ','
|
||
map keys terminated by ':'
|
||
lines terminated by '\n'
|
||
stored as textfile;
|
||
```
|
||
|
||
5. 加载数据。
|
||
|
||
```SQL
|
||
load data local inpath '/home/ubuntu/data/user_info/user_info.txt' overwrite into table dim_user_info;
|
||
```
|
||
|
||
或
|
||
|
||
```SQL
|
||
load data inpath '/user/root/user_info.txt' overwrite into table dim_user_info;
|
||
```
|
||
|
||
6. 创建分区表。
|
||
|
||
```SQL
|
||
create table if not exists fact_user_trade
|
||
(
|
||
user_name string,
|
||
piece int,
|
||
price double,
|
||
pay_amount double,
|
||
goods_category string,
|
||
pay_time bigint
|
||
)
|
||
partitioned by (dt string)
|
||
row format delimited fields terminated by '\t';
|
||
```
|
||
|
||
7. 提供分区数据。
|
||
|
||
```Shell
|
||
hdfs dfs -put /home/ubuntu/data/user_trade/* /user/hive/warehouse/eshop.db/fact_user_trade
|
||
```
|
||
|
||
8. 设置动态分区。
|
||
|
||
```SQL
|
||
set hive.exec.dynamic.partition=true;
|
||
set hive.exec.dynamic.partition.mode=nonstrict;
|
||
set hive.exec.max.dynamic.partitions=10000;
|
||
set hive.exec.max.dynamic.partitions.pernode=10000;
|
||
```
|
||
|
||
9. 修复分区。
|
||
|
||
```SQL
|
||
msck repair table fact_user_trade;
|
||
```
|
||
|
||
### 查询
|
||
|
||
#### 基本语法
|
||
|
||
```SQL
|
||
-- 查询北京女用户的姓名取前10个
|
||
select user_name from dim_user_info where city='beijing' and sex='female' limit 10;
|
||
|
||
-- 查询2019年3月24日购买了food类商品的用户名、购买数量和支付金额（不聚合）
|
||
select user_name, piece, pay_amount from fact_user_trade where dt='2019-03-24' and goods_category='food';
|
||
|
||
-- 统计用户 ELLA 在2018年的总支付金额和最近最远两次消费间隔天数
|
||
select sum(pay_amount) as total, datediff(max(from_unixtime(pay_time, 'yyyy-MM-dd')), min(from_unixtime(pay_time, 'yyyy-MM-dd'))) from fact_user_trade where year(dt)='2018' and user_name='ELLA';
|
||
```
|
||
|
||
#### group by
|
||
|
||
```SQL
|
||
-- 查询2019年1月到4月，每个品类有多少人购买，累计金额是多少
|
||
select goods_category, count(distinct user_name) as total_user, sum(pay_amount) as total_pay from fact_user_trade where dt between '2019-01-01' and '2019-04-30' group by goods_category;
|
||
```
|
||
|
||
```SQL
|
||
-- 查询2019年4月支付金额超过5万元的用户
|
||
select user_name, sum(pay_amount) as total from fact_user_trade where dt between '2019-04-01' and '2019-04-30' group by user_name having sum(pay_amount) > 50000;
|
||
```
|
||
|
||
```hive
|
||
-- 查询2018年购买的商品品类在两个以上的用户数
|
||
select count(tmp.user_name) from (select user_name, count(distinct goods_category) as total from fact_user_trade where year(dt)='2018' group by user_name having count(distinct goods_category)>2) tmp;
|
||
```
|
||
|
||
#### order by
|
||
|
||
```SQL
|
||
-- 查询2019年4月支付金额最多的用户前5名
|
||
select user_name, sum(pay_amount) as total from fact_user_trade where dt between '2019-04-01' and '2019-04-30' group by user_name order by total desc limit 5;
|
||
```
|
||
|
||
#### 常用函数
|
||
|
||
1. `from_unixtime`：将时间戳转换成日期
|
||
|
||
```hive
|
||
select from_unixtime(pay_time, 'yyyy-MM-dd hh:mm:ss') from fact_user_trade limit 10;
|
||
```
|
||
|
||
2. `unix_timestamp`：将日期转换成时间戳
|
||
|
||
3. `datediff`：计算两个日期的时间差
|
||
|
||
```Hive
|
||
-- 用户首次激活时间与设定参照时间的间隔
|
||
select user_name, datediff('2019-4-1', to_date(firstactivetime)) from dim_user_info limit 10;
|
||
```
|
||
|
||
4. `if`：根据条件返回不同的值
|
||
|
||
```Hive
|
||
-- 统计不同年龄段的用户数
|
||
select case when age < 20 then '20岁以下' when age < 30 then '30岁以下' when age < 40 then '40岁以下' else '40岁以上' end as age_seg, count(distinct user_id) as total from dim_user_info group by case when age < 20 then '20岁以下' when age < 30 then '30岁以下' when age < 40 then '40岁以下' else '40岁以上' end;
|
||
```
|
||
|
||
```Hive
|
||
-- 不同性别高级等用户数量
|
||
select sex, if(level > 5, '高', '低') as level_type, count(distinct user_id) as total from dim_user_info group by sex, if(level > 5, '高', '低');
|
||
```
|
||
|
||
5. `substr`：字符串取子串
|
||
|
||
```Hive
|
||
-- 统计每个月激活的新用户数
|
||
select substr(firstactivetime, 1, 7) as month, count(distinct user_id) as total from dim_user_info group by substr(firstactivetime, 1, 7);
|
||
```
|
||
|
||
6. `get_json_object`：从JSON字符串中取出指定的`key`对应的`value`，如：`get_json_object(info, '$.first_name')`。
|
||
|
||
```Hive
|
||
-- 统计不同手机品牌的用户数
|
||
select get_json_object(extra1, '$.phonebrand') as phone, count(distinct user_id) as total from user_info group by get_json_object(extra1, '$.phonebrand');
|
||
|
||
select extra2['phonebrand'] as phone, count(distinct user_id) as total from user_info group by extra2['phonebrand'];
|
||
```
|
||
|
||
> 说明：MySQL对应的函数名字叫`json_extract`。
|