nion和join是需要联合多张表时常见的关联词,具体概念我就不说了,想知道上网查就行,因为我也记不准确。
先说差别:union对两张表的操作是合并数据条数,等于是纵向的,要求是两张表字段必须是相同的(Schema of both sides of union shouldmatch.)。也就是说如果A表中有三条数据,B表中有两条数据,那么A union B就会有五条数据。说明一下union 和union all的差别,对于union如果存在相同的数据记录会被合并,而union all不会合并相同的数据记录,该有多少条记录就会有多少条记录。例如在mysql下执行以下语句:
select * fromtmp_libingxue_a;
name number
libingxue 1001
yuwen 1002
select * fromtmp_libingxue_b;
name number
libingxue 1001
feiyao 1003
select * fromtmp_libingxue_a union select * from tmp_libingxue_b;
libingxue 1001
yuwen 1002
feiyao 1003
select * fromtmp_libingxue_a union all select * from tmp_libingxue_b;
libingxue 1001
yuwen 1002
libingxue 1001
feiyao 1003
但是这样在hive里面是不能执行的,执行select * from tmp_libingxue_a union allselect * from tmp_libingxue_b;会failed,hive中union必须在子查询中进行。如
select * from(select * from tmp_yuwen_a union all select * from tmp_yuwen_b) t1;
注意,必须是union all,单独用union它会提示你缺少ALL,而且后面的t1必须写,你可以写成a或者b,但是一定要写,不写会出错。
而join则是偏于横向的联合,仅仅是偏向于,等下详细说明。join跟union比起来显得更宽松,对两个表的字段不做要求,没有限制条件的join等于两个表的笛卡尔乘积,所有join需要有限制条件来约束,经过限制的join就是横向的扩张了。对于满足限制条件的join会被提取出来,不满足的直接过滤掉。用法可以很灵活,下面有两个简单的例子:
select * from(select * from tmp_yuwen_a)t1 join (select * from tmp_yuwen_b) t2;
select * fromtmp_yuwen_a t1 join (select * from tmp_yuwen_b) t2;
left outer join和right outer join用法类似,区别就是left outer join会把左边表的字段全部选择出来,右边表的字段把符合条件的也选择出来,不满足的全部置空,也就是说以左边表为参照。right outer join同理以右边表为参照。这三个join之间的差别说过很多次,网上也有更详细的解释,不再赘述。
相同点:在某些特定的情况下,可以用join实现union all的功能,这种情况是有条件的,当出现这种情况的时候选择union all还是group by就可以看情况或者看两者的消耗而决定。sql虽然就在那么几个关键词,但变化多端、功能强大,只要能实现想要的功能,怎么用随便你。需求情况sql简单重现如下
drop tabletmp_libingxue_resource;
create externaltable if not exists tmp_libingxue_resource(
user_id string,
shop_id string,
auction_id string,
search_time string
)partitioned by(pt string)
row formatdelimited fields terminated by '\t'
lines terminatedby '\n'
stored assequencefile;
drop tabletmp_libingxue_result;
create externaltable if not exists tmp_libingxue_result(
user_id string,
shop_id string,
auction_id string,
search_time string
)partitioned by(pt string)
row formatdelimited fields terminated by '\t'
lines terminatedby '\n'
stored assequencefile;
insert overwritetable tmp_libingxue_result where(pt=20041104) select * fromtmp_libingxue_resource;
sudo -u taobaohadoop dfs -rmr /group/tbads/warehouse/tmp_libingxue_result/pt=20041104
sudo -u taobaohadoop jar /home/taobao/dataqa/framework/DailyReport.jarcom.alimama.loganalyzer.tool.SeqFileLoader tmp_libingxue_resource.txthdfs://v039182.sqa.cm4:54310/group/tbads/warehouse/tmp_libingxue_result/pt=20041104/part-00000
hive> select *from tmp_libingxue_resource;
OK
2001 0 11 10120041104
2002 0 11 102 20041104
hive> select *from tmp_libingxue_result;
OK
2001 0 12 10320041104
2002 0 12 10420041104
selectuser_id,shop_id,max(auction_id),max(search_time)
from
(select * fromtmp_libingxue_resource
union all
select * fromtmp_libingxue_result )t1
group byuser_id,shop_id;
2001 0 12 103
2002 0 12 104
selectt1.user_id,t1.shop_id,t2.auction_id,t2.search_time
from
(select * fromtmp_libingxue_resource) t1
join
(select * fromtmp_libingxue_result) t2
ont1.user_id=t2.user_id and t1.shop_id=t2.shop_id;
2001 0 12 103
2002 0 12 104