PGSQL表分区的实现方案和实践操作

表分区是解决一些因单表过大引用的性能问题的方式，比如某张表过大就会造成查询变慢，可能分区是一种解决方案。一般建议当单表大小超过内存就可以考虑表分区了。

目录

1，继承式分区，分为触发器（trigger）和规则（rule）两种方式

触发器的方式

规则的方式

2，声明式，分为list，range，hash三种方式

list方式

range方式

hash方式

3，分区表的维护

---

1，继承式分区，分为触发器（trigger）和规则（rule）两种方式

触发器的方式

1）创建表

CREATE TABLE "public"."track_info_rule_partition" (
    "id" serial,"object_type" int2 NOT NULL DEFAULT 0,"object_name" VARCHAR ( 255 ) COLLATE "pg_catalog"."default" NOT NULL DEFAULT '' :: CHARACTER VARYING,"object_value" VARCHAR ( 255 ) COLLATE "pg_catalog"."default" NOT NULL,"action_start_time" TIMESTAMP ( 0 ) NOT NULL,"action_end_time" TIMESTAMP ( 0 ) NOT NULL,"action_type" int2 NOT NULL,"action_geom" TEXT COLLATE "pg_catalog"."default" NOT NULL,"geom_name" VARCHAR ( 255 ) COLLATE "pg_catalog"."default" NOT NULL DEFAULT '' :: CHARACTER VARYING,"geom_number" VARCHAR ( 255 ) COLLATE "pg_catalog"."default" NOT NULL DEFAULT '' :: CHARACTER VARYING,"geom_address" VARCHAR ( 255 ) COLLATE "pg_catalog"."default" NOT NULL DEFAULT '' :: CHARACTER VARYING,"ctime" TIMESTAMP ( 0 ) NOT NULL DEFAULT now( ),"utime" TIMESTAMP ( 0 ) NOT NULL DEFAULT now( ),"is_del" int2 NOT NULL DEFAULT 0,"geom_station_start" VARCHAR ( 255 ) COLLATE "pg_catalog"."default" NOT NULL DEFAULT '' :: CHARACTER VARYING,"geom_station_end" VARCHAR ( 255 ) COLLATE "pg_catalog"."default" NOT NULL DEFAULT '' :: CHARACTER VARYING,CONSTRAINT "track_info_rule_partition_pkey" PRIMARY KEY ( "id" ) 
)

2）创建子表

DO $$ DECLARE
base TEXT;
sqlstring TEXT;
i INT;
BEGIN
        base = 'create table track_info_rule_partition_%s(check(action_start_time >= ''%s'' and action_start_time < ''%s'')) inherits (track_info_rule_partition)';
    FOR i IN 0..30
    loop
    sqlstring = format (
        base,to_char( '2022-05-15' :: DATE + ( i || ' day' ) :: INTERVAL,'YYYYMMDD' ),'2022-05-15' :: DATE + ( i || ' day' ) :: INTERVAL,'2022-05-15' :: DATE + ( i + 1 || ' day' ) :: INTERVAL 
    );
    EXECUTE sqlstring;
    
END loop;

END $$ LANGUAGE plpgsql;

 3）创建触发器函数

CREATE 
   OR REPLACE FUNCTION ins_or_upd_on_track_info_trigger_part ( ) RETURNS TRIGGER AS $$ DECLARE
   sqlstring TEXT;
BEGIN
      sqlstring = 'INSERT INTO track_info_trigger_partition_%s values($1.*)';
   sqlstring := format ( sqlstring,to_char( NEW.action_start_time,'YYYYMMDD' ) );
   EXECUTE sqlstring USING NEW;
   RETURN NULL;

END $$ LANGUAGE plpgsql;

4）创建触发器

CREATE TRIGGER tg_on_track_info_in_or_upd BEFORE INSERT ON track_info_trigger_partition FOR EACH ROW
EXECUTE FUNCTION ins_or_upd_on_track_info_trigger_part ( );

5）执行插入验证观察数据是否分布在子表和数据是否正常分布

规则的方式

1）创建大表

CREATE TABLE "public"."track_info_rule_partition" (
   "id" serial,CONSTRAINT "track_info_rule_partition_pkey" PRIMARY KEY ( "id" ) 
)

2）创建子表

DO $$ DECLARE
base TEXT;
sqlstring TEXT;
i INT;
BEGIN
    base = 'create table track_info_rule_partition_%s(check(action_start_time >= ''%s'' and action_start_time < ''%s'')) inherits (track_info_rule_partition)';
    FOR i IN 0..30
    loop
    sqlstring = format (
        base,'2022-05-15' :: DATE + ( i + 1 || ' day' ) :: INTERVAL 
    );
    EXECUTE sqlstring;
    
END loop;

END $$ LANGUAGE plpgsql;

3)创建规则

DO $$ DECLARE
base TEXT;
sqlstring TEXT;
i INT;
BEGIN
        base = 'CREATE OR REPLACE RULE track_info_rule_partition_%s AS ON INSERT TO track_info_rule_partition
    WHERE action_start_time >= ''%s'' and action_start_time < ''%s''
    DO INSTEAD
    INSERT INTO track_info_rule_partition_%s VALUES(NEW.*)';
    FOR i IN 0..30
    loop
    sqlstring = format (
        base,'2022-05-15' :: DATE + ( i + 1 || ' day' ) :: INTERVAL,'YYYYMMDD' )
        
    );
    EXECUTE sqlstring;
    
END loop;

END $$ LANGUAGE plpgsql;

4）执行插入验证观察数据是否分布在子表和数据是否正常分布

2，声明式，分为list，range，hash三种方式

list方式

1）创建list分区父表

CREATE TABLE "public"."track_info_list_partition" (
   "id" serial,CONSTRAINT "track_info_list_partition_pkey" PRIMARY KEY ( "id","action_type" ) 
) PARTITION BY list ( action_type );

2）创建分区子表

do                                                                    
$$
declare base text; sqlstring text; i int;
begin
base = 'create table track_info_list_partition_%s partition of track_info_list_partition for values in (''%s'')';
for i in 0..9 loop
    sqlstring = format(base,i,i);
    execute sqlstring;
end loop;                                          
end     
$$language plpgsql;

3）执行插入验证观察数据是否分布在子表和数据是否正常分布

range方式

1）创建大表

CREATE TABLE "public"."track_info_range_partition" (
   "id" serial,CONSTRAINT "track_info_range_partition_pkey" PRIMARY KEY ( "id","action_start_time" ) 
) PARTITION BY range ( action_start_time );

2）创建分区子表

DO
   $$ DECLARE
   base TEXT;
date_start TEXT = '2022-05-15';
sqlstring TEXT;
i INT;
BEGIN
      base = 'create table track_info_range_partition_%s partition of track_info_range_partition for values from (''%s'') to (''%s'')';
   FOR i IN 0..30
   loop
   sqlstring = format (
      base,to_char( date_start :: timestamptz + ( i || ' day' ) :: INTERVAL,date_start :: timestamptz + ( i || ' day' ) :: INTERVAL,date_start :: timestamptz + ( i + 1 || ' day' ) :: INTERVAL 
   );
   EXECUTE sqlstring;
   
END loop;

END $$ LANGUAGE plpgsql;

3）执行插入验证观察数据是否分布在子表和数据是否正常分布

hash方式

1）创建大表

CREATE TABLE "public"."track_info_hash_partition" (
   "id" serial,CONSTRAINT "track_info_hash_partition_pkey" PRIMARY KEY ( "id","geom_name" ) 
) PARTITION BY hash ( geom_name );

2）创建分区子表

DO
   $$ DECLARE
   base TEXT;
sqlstring TEXT;
i INT;
BEGIN
      base = 'create table track_info_hash_partition_%s partition of track_info_hash_partition for values with(modulus 10,remainder %s)';
   FOR i IN 0..9
   loop
   sqlstring = format ( base,i );
   EXECUTE sqlstring;
   
END loop;

END $$ LANGUAGE plpgsql;

3）执行插入验证观察数据是否分布在子表和数据是否正常分布

3，分区表的维护

继承式分区表的维护，主要为 INHERIT 与 NO INHERIT。语法如下:

ALTER TABLE track_info_trigger_partition_20220520 NO INHERIT track_info_trigger_partition;

ALTER TABLE track_info_trigger_partition_20220520 INHERIT track_info_trigger_partition;

NO INHERIT的情况下，不可以通过父表对子表进行CRUD，INHERIT之后可以正常的CRUD。

声明式分区表的维护，主要为 ATTACH 与 DETACH。语法如下:

ALTER TABLE track_info_list_partition DETACH PARTITION track_info_list_partition_4;

ALTER TABLE track_info_list_partition ATTACH PARTITION track_info_list_partition_4 FOR VALUES in (4)

DETACH的情况下，不可以通过父表对子表进行CRUD，ATTACH之后可以正常的CRUD。