PostgreSQL 多级分区

CREATE TABLE range_list (a int,b timestamp) PARTITION BY RANGE (b);

CREATE TABLE range_pa1 PARTITION OF range_list FOR VALUES from ('2000-01-01') TO ('2010-01-01')  PARTITION BY RANGE (a);
CREATE TABLE range_pa2 PARTITION OF range_list FOR VALUES from ('2010-01-01') TO ('2020-01-01')  PARTITION BY RANGE (a);

CREATE TABLE range_list_2000_2010_1_10  PARTITION OF range_pa1 FOR VALUES from (1) TO (10);
CREATE TABLE range_list_2000_2010_10_20 PARTITION OF range_pa1 FOR VALUES from (11) TO (20);

CREATE TABLE range_list_2010_2020_1_10  PARTITION OF range_pa2 FOR VALUES from (1) TO (10);
CREATE TABLE range_list_2010_2020_10_20 PARTITION OF range_pa2 FOR VALUES from (11) TO (20);

exampledb=> \d
                 List of relations
 Schema |        Name        |   Type   |  Owner
--------+--------------------+----------+----------
 public | range_list1_10     | table    | dbuser
 public | range_list1_20     | table    | dbuser
 public | range_list2_10     | table    | dbuser
 public | range_list2_20     | table    | dbuser

exampledb=> insert into range_list values(1,'2005-01-05 5:05');
INSERT 0 1

exampledb=> explain select * from range_list where a=1 and b='2005-01-05 5:05';
               QUERY PLAN
----------------------------------------------------------------------------------------
 Append  (cost=0.00..40.60 rows=1 width=12)
   ->  Seq Scan on range_list1_10  (cost=0.00..40.60 rows=1 width=12)
         Filter: ((a = 1) AND (b = '2005-01-05 05:05:00'::timestamp without time zone))
(3 rows)

Time: 1.425 ms

 

 一般开发中，大多需要用户ID和时间区间进行查询数据，可以使用此方法进行二位分区。