Order by和Sort by的区别?
Order by
Order by会引发全局排序,有可能会导致任务失败
select * from baidu_click order by click desc;
Sort by
distribute by + sort by就是该替代方案,被distribute by设定的字段为KEY,数据会被HASH分发到不同的reducer机器上,然后sort by会对同一个reducer机器上的每组数据进行局部排序。
select * from baidu_click distribute by product_line sort by click desc;