本节简单介绍了PostgreSQL中和distinct和group by.
通常来说,获取唯一值,既可以用distinct也可以用group by,但在存在主键时,group by会做相应的优化,把多个分组键规约为主键.
没有主键的情况

[pg12@localhost ~]$ psql
Expanded display is used automatically.
psql (12.2)
Type "help" for help.
[local:/data/run/pg12]:5120 pg12@testdb=# create table tbl1 (id int,c1 text,c2 int,c3 varchar);
CREATE TABLE
[local:/data/run/pg12]:5120 pg12@testdb=# insert into tbl1 (id,c1,c2,c3) select x,x||'c1',x,x||'c3' from generate_series(1,100000) as x;
INSERT 0 100000
[local:/data/run/pg12]:5120 pg12@testdb=# explain select distinct id,c1,c2,c3 from tbl1;
                            QUERY PLAN                            
------------------------------------------------------------------
 HashAggregate  (cost=1668.94..1720.54 rows=5160 width=72)
   Group Key: id, c1, c2, c3
   ->  Seq Scan on tbl1  (cost=0.00..1152.97 rows=51597 width=72)
(3 rows)
[local:/data/run/pg12]:5120 pg12@testdb=# explain select id,c1,c2,c3 from tbl1 group by id,c1,c2,c3;
                            QUERY PLAN                            
------------------------------------------------------------------
 HashAggregate  (cost=1668.94..1720.54 rows=5160 width=72)
   Group Key: id, c1, c2, c3
   ->  Seq Scan on tbl1  (cost=0.00..1152.97 rows=51597 width=72)
(3 rows)

存在主键的情况

[local:/data/run/pg12]:5120 pg12@testdb=# alter table tbl1 add primary key(id);
'ALTER TABLE
[local:/data/run/pg12]:5120 pg12@testdb=# explain select distinct id,c1,c2,c3 from tbl1;
                               QUERY PLAN                                
-------------------------------------------------------------------------
 Unique  (cost=14043.82..15293.82 rows=100000 width=72)
   ->  Sort  (cost=14043.82..14293.82 rows=100000 width=72)
         Sort Key: id, c1, c2, c3
         ->  Seq Scan on tbl1  (cost=0.00..1637.00 rows=100000 width=72)
(4 rows)
[local:/data/run/pg12]:5120 pg12@testdb=# explain select id,c1,c2,c3 from tbl1 group by id,c1,c2,c3;
                                     QUERY PLAN                                      
-------------------------------------------------------------------------------------
 Group  (cost=0.29..5402.29 rows=100000 width=72)
   Group Key: id
   ->  Index Scan using tbl1_pkey on tbl1  (cost=0.29..5152.29 rows=100000 width=72)
(3 rows)
[local:/data/run/pg12]:5120 pg12@testdb=#

在存在主键的情况下,使用group by时,分组键只需要主键即可.