Hive实现增量更新：full outer join 合并 + insert overwrite（阿里巴巴大数据实践中阿里平台使用方案）

首先感谢原文作者给我的启发，我加以修改。

https://blog.csdn.net/magicharvey/article/details/20692829https://blog.csdn.net/CLKTOY/article/details/120306523

前言
我一直有个困惑，数仓日增量表，历史数据修改了除了拉链表还能怎么办？
因为我们同步过来的的增量数据不止有新增的数据还有修改的数据，这里我们不考虑拉链表，只存放当前最新的状态。
后来看了1和2，惭愧惭愧。
说实话我也看了阿里巴巴中台建设这本书，我也看到了p42页这一块full outer join 合并 + insert overwrite，但是我嫌复杂没仔细看，我觉得书在胡言乱语，不可能实现吧。
当我们增量同步进来的数据是和我们的历史数据进行合并的，这个时候我们就可以使用full join 。
我做了一定的修改，直接进入正题吧。

原有学生表student

create table student (
id string,
age string,
name string,
dt string
);

insert into table student values
("1","11","zhao","20140101")
,("2","22","qian","20140102")
,("3","33","sun","20140103")
,("4","44","li","20140104");

select * from student;

现有最新的学生表student_temp

数据较原有student的表，数据有新增也有修改

create table student_temp(
id string,
age string,
name string,
dt string
);

insert into table student_temp values
("1","11","zhao","20140101")   -- 原本数据
,("2","999","test","20220323") -- 改动
,("3","999","test","20220323") -- 改动
,("4","44","li","20140104")    -- 原本数据
,("5","55","wang","20140105")  -- 新增数据
;

select * from student_temp;

查看full outer join效果

select * from student_temp a full outer join student b on a.id = b.id;

开始更新

没有则新增，有则覆盖

SELECt  student_temp.id
        ,coalesce(student_temp.age,student.age) as age 
        ,student_temp.name
        ,coalesce(student_temp.dt,student.dt) as dt
FROM    student_temp
FULL OUTER JOIN student
ON      student_temp.id = student.id
;