postgresql----表分区

–下面的描述不记得在哪里抄来的了?!

表分区就是把逻辑上一个大表分割成物理上的多个小块,表分区可提供如下若干好处:

1.某些类型的查询性能可以得到极大提升。

2.更新的性能可以得到提升,因为表的每块索引要比整个数据集上的索引要小,如果索引不能全部放在内存里,那么在索引上的读写都会产生磁盘访问。

3.批量删除可以用简单的删除某个分区

4.将很少使用的数据移动到便宜的慢一些的存储介质上。

示例1. 

1.创建主表

create table tbl_inherits_test
(
    a int,
    b timestamp without time zone
);
create index idx_tbl_inherits_test_b on tbl_inherits_test using btree (b);

 

2.创建触发器函数,在INSERT父表时根据时间字段b写入时间b的分表,如果分表b不存在,则创建分表b,然后再INSERT分表

create or replace function f_insert_tbl_inherits_test() returns trigger as
$body$
declare tablename varchar(32) default '';
begin
    tablename='tbl_inherits_test_'||to_char(NEW.b,'YYYY_MM_DD');
   
    execute 'insert into '||tablename||'(a,b) values('||NEW.a||','''||NEW.b||''')';
    return null;
    EXCEPTION
        when undefined_table then
        execute 'create table '||tablename||'() inherits (tbl_inherits_test)';
        execute 'create index idx_'||tablename||'_b on '||tablename||' using btree(b)';
        execute 'insert into '||tablename||'(a,b) values('||NEW.a||','''||NEW.b||''')';
    return null;    
end;
$body$
language plpgsql; 

 

3.创建触发器,当INSERT主表时执行触发器函数

create trigger trg_insert_tbl_inherits_test before insert on tbl_inherits_test for each row execute procedure f_insert_tbl_inherits_test();

 

4.向主表写数据验证结果

test=#insert into tbl_inherits_test(a,b) values(1,'2016-06-20 17:40:21');
test=# \d+ tbl_inherits_test 
                            Table "public.tbl_inherits_test"
 Column |            Type             | Modifiers | Storage | Stats target | Description 
--------+-----------------------------+-----------+---------+--------------+-------------
 a      | integer                     |           | plain   |              | 
 b      | timestamp without time zone |           | plain   |              | 
Indexes:
    "idx_tbl_inherits_test_b" btree (b)
Triggers:
    trg_insert_tbl_inherits_test BEFORE INSERT ON tbl_inherits_test FOR EACH ROW EXECUTE PROCEDURE f_insert_tbl_inherits_test()
Child tables: tbl_inherits_test_2016_06_20

5.结果显示INSERT主表时会根据INSERT的数据b(2016-06-20 17:40:21)自动创建一个分表tbl_inherits_test_2016_06_20,再写入几条数据,查看结果

test=# insert into tbl_inherits_test(a,b) values (2,'2016-06-20 08:08:08'),(3,'2016-06-21 19:00:00');
INSERT 0 0
test=# \d+ tbl_inherits_test
                            Table "public.tbl_inherits_test"
 Column |            Type             | Modifiers | Storage | Stats target | Description 
--------+-----------------------------+-----------+---------+--------------+-------------
 a      | integer                     |           | plain   |              | 
 b      | timestamp without time zone |           | plain   |              | 
Indexes:
    "idx_tbl_inherits_test_b" btree (b)
Triggers:
    trg_insert_tbl_inherits_test BEFORE INSERT ON tbl_inherits_test FOR EACH ROW EXECUTE PROCEDURE f_insert_tbl_inherits_test()
Child tables: tbl_inherits_test_2016_06_20,
              tbl_inherits_test_2016_06_21

 

6.分别查询主表和分表的数据,直接查询主表会查询到所有分表的数据,但是使用only查询主表发现,主表中并没有数据(因为触发器函数中返回的是null)

test=# select * from tbl_inherits_test_2016_06_20 ;
 a |          b          
---+---------------------
 1 | 2016-06-20 17:40:21
 2 | 2016-06-20 08:08:08
(2 rows)

test=# select * from tbl_inherits_test_2016_06_21 ;
 a |          b          
---+---------------------
 3 | 2016-06-21 19:00:00
(1 row)

test=# 
test=# select * from tbl_inherits_test ;
 a |          b          
---+---------------------
 1 | 2016-06-20 17:40:21
 2 | 2016-06-20 08:08:08
 3 | 2016-06-21 19:00:00
(3 rows)

test=# select * from only tbl_inherits_test ;
 a | b 
---+---
(0 rows)

 

    原文作者:PostgreSQL
    原文地址: https://www.cnblogs.com/alianbog/p/5605234.html
    本文转自网络文章,转载此文章仅为分享知识,如有侵权,请联系博主进行删除。
点赞