PostgreSQL 10中如何使用分区表

本篇文章给大家分享的是有关PostgreSQL 10中如何使用分区表，小编觉得挺实用的，因此分享给大家学习，希望大家阅读完这篇文章后可以有所收获，话不多说，跟着小编一起来看看吧。

分区介绍

PostgreSQL的分区需要先建立主表，然后再建立子表，使用继承的特性，但不需要手动写触发器/规则了，目前支持range、list分区，10 正式版发布时不知道会不会有其他的，后面我会介绍我基于10 Beta2添加的hash分区。

range分区

分区语法：

postgres=# create table r (r_id int, r_name name, r_date date) partition by range (r_id); CREATE TABLE postgres=# create table r1 partition of r for values from (1) to (10); CREATE TABLE postgres=# create table r2 partition of r for values from (10) to (20); CREATE TABLE postgres=# create table r3 partition of r for values from (20) to (30); CREATE TABLE postgres=# insert into r select id, md5(random()::text), now() + (id||'day')::interval from generate_series(1, 29) t(id); INSERT 0 29 postgres=# select *, tableoid::regclass from r;  r_id |  r_name  |   r_date   | tableoid  ------+----------------------------------+------------+----------     1 | 1d0d0680930198d2962b3b5f9cf82083 | 2017-08-09 | r1     2 | 47ba81de41d71bd51b18c7861a594bdf | 2017-08-10 | r1     3 | 820b0b1affe3bf0e5705aee3e77b0b29 | 2017-08-11 | r1     4 | 0cc06451bd0652d2583a733374d787b3 | 2017-08-12 | r1     5 | 642108381b2fc203b830f1215a0d7c6a | 2017-08-13 | r1     6 | 57e3869b2ab8ee1c0bca96b1cf022a5d | 2017-08-14 | r1     7 | 5357fa6de3c1c559edb78cddb4eae902 | 2017-08-15 | r1     8 | 6ea5a7dba4dfc6c81ca5932be86a9341 | 2017-08-16 | r1     9 | d3d4dcb9dc48e0629042ede7ed9c7a33 | 2017-08-17 | r1    10 | 248d6f3e072c6c137a3402d11fc5b1d7 | 2017-08-18 | r2    11 | ae3a671045ded43260bc4d0bbcb7e428 | 2017-08-19 | r2    12 | acdc89bb326d9f0caaeeb86bfeac3a76 | 2017-08-20 | r2    13 | 147b6e975d7299db66e170874b913b25 | 2017-08-21 | r2    14 | 6041a6b84b1af615bdb34a5926d72a33 | 2017-08-22 | r2    15 | 3d96e08395af120dd36e10a0252ce29c | 2017-08-23 | r2    16 | 5e613d10c9cac126453413ddfc17c210 | 2017-08-24 | r2    17 | e92fc34d180be652e72a63b92d327f1b | 2017-08-25 | r2    18 | 3109c4e8f4da701721151df11a4d266f | 2017-08-26 | r2    19 | 35ba5892f3b88aa3254445fbf5267eea | 2017-08-27 | r2    20 | c92d1df47257784bb11d7bfbb52b5710 | 2017-08-28 | r3    21 | d076a5498d17ade8f317bf47cfa322c3 | 2017-08-29 | r3    22 | a66c2e83f1e54e1392964ed71d5b8e20 | 2017-08-30 | r3    23 | 6a94df0f08921728aa0af9455d05c9f8 | 2017-08-31 | r3    24 | 248c46d80b926c66c093c500f309614d | 2017-09-01 | r3    25 | 4da3be147fd1831e8605fc400e7a7503 | 2017-09-02 | r3    26 | 3029d7e22b7c963e8983200a93894669 | 2017-09-03 | r3    27 | 720d6d04249e9f3595a19cf59f075332 | 2017-09-04 | r3    28 | 95b5e5492591c38ddd864d83265e26c4 | 2017-09-05 | r3    29 | 2628c14bd3f67699ab0411b6fd402460 | 2017-09-06 | r3 (29 rows)  postgres=# explain select * from r where id = 20; ERROR:  column "id" does not exist LINE 1: explain select * from r where id = 20;   ^ postgres=# explain select * from r where r_id = 20; QUERY PLAN ----------------------------------------------------------  Append  (cost=0.00..20.12 rows=4 width=72)    ->  Seq Scan on r3  (cost=0.00..20.12 rows=4 width=72)  Filter: (r_id = 20) (3 rows)  postgres=# set constraint_exclusion = off; SET postgres=# explain select * from r where r_id = 20; QUERY PLAN ----------------------------------------------------------  Append  (cost=0.00..60.38 rows=12 width=72)    ->  Seq Scan on r1  (cost=0.00..20.12 rows=4 width=72)  Filter: (r_id = 20)    ->  Seq Scan on r2  (cost=0.00..20.12 rows=4 width=72)  Filter: (r_id = 20)    ->  Seq Scan on r3  (cost=0.00..20.12 rows=4 width=72)  Filter: (r_id = 20) (7 rows)  postgres=#   postgres=# create index on r1 (r_id); CREATE INDEX postgres=# explain select * from r where r_id = 5; QUERY PLAN ----------------------------------------------------------------------------------  Append  (cost=5.53..25.54 rows=161 width=72)    ->  Bitmap Heap Scan on r1  (cost=5.53..25.54 rows=161 width=72)  Recheck Cond: (r_id = 5)  ->  Bitmap Index Scan on r1_r_id_idx  (cost=0.00..5.48 rows=161 width=0)    Index Cond: (r_id = 5) (5 rows)  postgres=# \d+ r*  Table "public.r"  Column |  Type   | Collation | Nullable | Default | Storage | Stats target | Description  --------+---------+-----------+----------+---------+---------+--------------+-------------  r_id   | integer |   |  | | plain   |  |   r_name | name|   |  | | plain   |  |   r_date | date|   |  | | plain   |  |  Partition key: RANGE (r_id) Partitions: r1 FOR VALUES FROM (1) TO (10), r2 FOR VALUES FROM (10) TO (20), r3 FOR VALUES FROM (20) TO (30)  Table "public.r1"  Column |  Type   | Collation | Nullable | Default | Storage | Stats target | Description  --------+---------+-----------+----------+---------+---------+--------------+-------------  r_id   | integer |   |  | | plain   |  |   r_name | name|   |  | | plain   |  |   r_date | date|   |  | | plain   |  |  Partition of: r FOR VALUES FROM (1) TO (10) Partition constraint: ((r_id IS NOT NULL) AND (r_id >= 1) AND (r_id < 10)) Indexes: "r1_r_id_idx" btree (r_id)     Index "public.r1_r_id_idx"  Column |  Type   | Definition | Storage  --------+---------+------------+---------  r_id   | integer | r_id   | plain btree, for table "public.r1"  Table "public.r2"  Column |  Type   | Collation | Nullable | Default | Storage | Stats target | Description  --------+---------+-----------+----------+---------+---------+--------------+-------------  r_id   | integer |   |  | | plain   |  |   r_name | name|   |  | | plain   |  |   r_date | date|   |  | | plain   |  |  Partition of: r FOR VALUES FROM (10) TO (20) Partition constraint: ((r_id IS NOT NULL) AND (r_id >= 10) AND (r_id < 20))  Table "public.r3"  Column |  Type   | Collation | Nullable | Default | Storage | Stats target | Description  --------+---------+-----------+----------+---------+---------+--------------+-------------  r_id   | integer |   |  | | plain   |  |   r_name | name|   |  | | plain   |  |   r_date | date|   |  | | plain   |  |  Partition of: r FOR VALUES FROM (20) TO (30) Partition constraint: ((r_id IS NOT NULL) AND (r_id >= 20) AND (r_id < 30))

说明：

• 创建分区时必须指定主表

• 分区表和主表列数量、定义必须完全一致

• 分区表的列可以单独添加约束、索引

• 向主表插入数据，自动插入到对应分区，如果找不到对应分区，抛出错误

• range分区范围>=最小值、<***值

• 修改主表字段名、字段类型，会自动修改所有分区

• truncate主表会清除所有分区表数据

• drop主表会把所有子表一起drop

• \d、\d+ 可查看分区表详细定义

在PostgreSQL10的分区表功能中， 范围分区的KEY支持由多个字段组成，多列组成的KEY可看做是范围分区表的组合约束。

[postgres@localhost bin]$ ./psql  psql (10beta2) Type "help" for help.  postgres=# create table r(a int, b int) partition by range (a, b); CREATE TABLE postgres=# create table r1 partition of r for values from (1, 60) to (10, 80); CREATE TABLE postgres=# create table r2 partition of r for values from (10, 80) to (20, 60); CREATE TABLE  postgres=# \d+ r*                                      Table "public.r"  Column |  Type   | Collation | Nullable | Default | Storage | Stats target | Description  --------+---------+-----------+----------+---------+---------+--------------+-------------  a      | integer |           |          |         | plain   |              |   b      | integer |           |          |         | plain   |              |  Partition key: RANGE (a, b) Partitions: r1 FOR VALUES FROM (1, 60) TO (10, 80),             r2 FOR VALUES FROM (10, 80) TO (20, 60)                                      Table "public.r1"  Column |  Type   | Collation | Nullable | Default | Storage | Stats target | Description  --------+---------+-----------+----------+---------+---------+--------------+-------------  a      | integer |           |          |         | plain   |              |   b      | integer |           |          |         | plain   |              |  Partition of: r FOR VALUES FROM (1, 60) TO (10, 80) Partition constraint: ((a IS NOT NULL) AND (b IS NOT NULL) AND ((a > 1) OR ((a = 1) AND (b >= 60))) AND ((a < 10) OR ((a = 10) AND (b < 80))))                                      Table "public.r2"  Column |  Type   | Collation | Nullable | Default | Storage | Stats target | Description  --------+---------+-----------+----------+---------+---------+--------------+-------------  a      | integer |           |          |         | plain   |              |   b      | integer |           |          |         | plain   |              |  Partition of: r FOR VALUES FROM (10, 80) TO (20, 60) Partition constraint: ((a IS NOT NULL) AND (b IS NOT NULL) AND ((a > 10) OR ((a = 10) AND (b >= 80))) AND ((a < 20) OR ((a = 20) AND (b < 60))))  postgres=# insert into r values (10, 70); INSERT 0 1 postgres=# insert into r values (10, 80); INSERT 0 1 postgres=# insert into r values (10, 90); INSERT 0 1 postgres=# select tableoid::regclass, * from r;  tableoid | a  | b   ----------+----+----  r1       | 10 | 70  r2       | 10 | 80  r2       | 10 | 90 (7 rows)  postgres=#

这里需要注意它的分区约束，from (10, 80) to (20, 60)，最初还以为是有bug，其实不是，Partition constraint: ((a IS NOT NULL) AND (b IS NOT NULL) AND ((a > 10) OR ((a = 10) AND (b >= 80))) AND ((a < 20) OR ((a = 20) AND (b < 60))))

list分区

语法：

postgres=# create table l (l_id int, l_name name, l_date date) partition by list (l_id); CREATE TABLE postgres=# create table l1 partition of l for values in (1); CREATE TABLE postgres=# create table l2 partition of l for values in (2); CREATE TABLE postgres=# create table l3 partition of l for values in (3); CREATE TABLE postgres=# create table l4 partition of l for values in (4); CREATE TABLE postgres=# insert into l select id, md5(random()::text), now() + (id||'day')::interval from generate_series(1, 5) t(id); ERROR:  no partition of relation "l" found for row DETAIL:  Partition key of the failing row contains (l_id) = (5). postgres=# insert into l select id, md5(random()::text), now() + (id||'day')::interval from generate_series(1, 4) t(id); INSERT 0 4 postgres=# explain select * from l where l_id = 2; QUERY PLAN ----------------------------------------------------------  Append  (cost=0.00..20.12 rows=4 width=72)    ->  Seq Scan on l2  (cost=0.00..20.12 rows=4 width=72)  Filter: (l_id = 2) (3 rows)  postgres=#

hash分区

语法：

[postgres@localhost bin]$ ./psql yonj1e psql (10beta2) Type "help" for help.  yonj1e=# create table h (h_id int, h_name name, h_date date) partition by hash(h_id); CREATE TABLE yonj1e=# create table h2 partition of h; CREATE TABLE yonj1e=# create table h3 partition of h; CREATE TABLE yonj1e=# create table h4 partition of h; CREATE TABLE yonj1e=# create table h5 partition of h; CREATE TABLE yonj1e=# insert into h select id, md5(random()::text), now() + (id||'day')::interval from generate_series(1, 50) t(id); INSERT 0 50 yonj1e=# select *,tableoid::regclass from h;  h_id |  h_name  |   h_date   | tableoid  ------+----------------------------------+------------+----------     5 | 21fe9a616ce20868769904bbda56aa3e | 2017-08-14 | h2     6 | 8fa0f42bf4239c05c1cd46a814f71eaa | 2017-08-15 | h2     8 | 858e324311506fb5c5000a4741b9af3c | 2017-08-17 | h2    12 | ef4ce7a0f6168605a7c243a709f28bc3 | 2017-08-21 | h2    13 | 2273522a7b3c286e214a8f57e010568e | 2017-08-22 | h2    17 | 8bca453f60f13278d3a02149b30394d2 | 2017-08-26 | h2    19 | 0c2f14a6a8e675341b4e7bdb6ed161de | 2017-08-28 | h2    23 | f10fff43558393b577d417127bf6a163 | 2017-09-01 | h2    26 | 1dd0851728458b67a053d500bbb837ae | 2017-09-04 | h2    28 | 00f67b8636b4d225d3b62bcca9c6d527 | 2017-09-06 | h2    40 | d3a217d39b6808ff5e37ed3977513e05 | 2017-09-18 | h2    41 | 0f4c765d809c3db3fa608e986aed1247 | 2017-09-19 | h2    42 | 022ff983201352092d5d7cb735e9f531 | 2017-09-20 | h2    44 | c3dba31501b3625aac7f3d4f41512855 | 2017-09-22 | h2    49 | 21c697e92f936982840b928e03151204 | 2017-09-27 | h2    11 | 625ad3b0c9d40f7cae26be84a7ae054d | 2017-08-20 | h3    14 | 7ee39c8df46d7ec61923dffe6f58ec07 | 2017-08-23 | h3    22 | 77c9230f9eeeb9faaa5c30ff518bbf60 | 2017-08-31 | h3    29 | 5a6e0895b2477026bcfa4996650797a8 | 2017-09-07 | h3    31 | 37d84c0c0956df75407e0bfc67a782ed | 2017-09-09 | h3    34 | f43f07545fba020b47c84952c6af6cc7 | 2017-09-12 | h3    35 | 2fa08f1311c20ac45a6726bfbc8a4f05 | 2017-09-13 | h3    36 | d01940a876b86c2de8d67a37813ab89d | 2017-09-14 | h3    38 | f21e401cb38c6d625b264f53fb59fb8b | 2017-09-16 | h3    43 | c42fcb14c2d5e5c067db8231d502daae | 2017-09-21 | h3    45 | cc5670020ba35ae2c324dd33a6efff98 | 2017-09-23 | h3     1 | 2881b2aadddd2dfef14477369a107319 | 2017-08-10 | h4     2 | e22e0bba2716e6d969a62502d34fc518 | 2017-08-11 | h4     9 | a933091df7f51f5b0b6ab43816e5d765 | 2017-08-18 | h4    15 | 92ee6dc6670ea8e02e746ee508b51022 | 2017-08-24 | h4    21 | 67b7140105e81730f364ee9de195e0a0 | 2017-08-30 | h4    46 | f4c47b9e055f6c732dff55cb4fd152b3 | 2017-09-24 | h4    50 | 1b419faea293d3edab2cfb7f8efb55f8 | 2017-09-28 | h4     3 | 6ea55fe46f2119f084edd66abe486d91 | 2017-08-12 | h5     4 | 094b16011f0e9b34c878caa7d95e067f | 2017-08-13 | h5     7 | c916c264c77ba90b3463143b9513bc15 | 2017-08-16 | h5    10 | d333ce15f3a8d01a39df9ae317bf64b7 | 2017-08-19 | h5    16 | b136e8466bbafc58f917e14919b1edf1 | 2017-08-25 | h5    18 | 6c3fb7b3e473f793575407299006e6a3 | 2017-08-27 | h5    20 | bac12655b2d54855c58d19c9facc1579 | 2017-08-29 | h5    24 | ebcafb42d26654eff04bb5f8b35fdd69 | 2017-09-02 | h5    25 | 494e25facb9fe46e00037c716e2052e7 | 2017-09-03 | h5    27 | 961b1728893e7f6d46ed827ee9b4809e | 2017-09-05 | h5    30 | ebf840a36af46cc3dd8f29c94013cb71 | 2017-09-08 | h5    32 | 7c5083fed360079bcbc23c6ee803a4d6 | 2017-09-10 | h5    33 | 707e98eac5ba349c80df6f9d8f062676 | 2017-09-11 | h5    37 | e1cb546e66cd45b00441493100fb7752 | 2017-09-15 | h5    39 | 0bfa7e7ccb00a477add00df4d0327c52 | 2017-09-17 | h5    47 | d894969c3cdbf00fe7dce9683c6f9a17 | 2017-09-25 | h5    48 | 09ca86c5e5bd87ba786533f34c4ebf25 | 2017-09-26 | h5 (50 rows)  yonj1e=# explain select * from h where h_id = 20; QUERY PLAN ----------------------------------------------------------  Append  (cost=0.00..20.12 rows=4 width=72)    ->  Seq Scan on h5  (cost=0.00..20.12 rows=4 width=72)  Filter: (h_id = 20) (3 rows) yonj1e=# \d+ h*  Table "public.h"  Column |  Type   | Collation | Nullable | Default | Storage | Stats target | Description  --------+---------+-----------+----------+---------+---------+--------------+-------------  h_id   | integer |   |  | | plain   |  |   h_name | name|   |  | | plain   |  |   h_date | date|   |  | | plain   |  |  Partition key: HASH (h_id) Partitions: h2 SERIAL NUMBER 0, h3 SERIAL NUMBER 1, h4 SERIAL NUMBER 2, h5 SERIAL NUMBER 3  Table "public.h2"  Column |  Type   | Collation | Nullable | Default | Storage | Stats target | Description  --------+---------+-----------+----------+---------+---------+--------------+-------------  h_id   | integer |   |  | | plain   |  |   h_name | name|   |  | | plain   |  |   h_date | date|   |  | | plain   |  |  Partition of: h SERIAL NUMBER 0 Partition constraint: (h_id IS NOT NULL)  Table "public.h3"  Column |  Type   | Collation | Nullable | Default | Storage | Stats target | Description  --------+---------+-----------+----------+---------+---------+--------------+-------------  h_id   | integer |   |  | | plain   |  |   h_name | name|   |  | | plain   |  |   h_date | date|   |  | | plain   |  |  Partition of: h SERIAL NUMBER 1 Partition constraint: (h_id IS NOT NULL)  Table "public.h4"  Column |  Type   | Collation | Nullable | Default | Storage | Stats target | Description  --------+---------+-----------+----------+---------+---------+--------------+-------------  h_id   | integer |   |  | | plain   |  |   h_name | name|   |  | | plain   |  |   h_date | date|   |  | | plain   |  |  Partition of: h SERIAL NUMBER 2 Partition constraint: (h_id IS NOT NULL)  Table "public.h5"  Column |  Type   | Collation | Nullable | Default | Storage | Stats target | Description  --------+---------+-----------+----------+---------+---------+--------------+-------------  h_id   | integer |   |  | | plain   |  |   h_name | name|   |  | | plain   |  |   h_date | date|   |  | | plain   |  |  Partition of: h SERIAL NUMBER 3 Partition constraint: (h_id IS NOT NULL)  yonj1e=#

HASH分区语法还不支持，以后或许会支持。

以上就是PostgreSQL 10中如何使用分区表，小编相信有部分知识点可能是我们日常工作会见到或用到的。希望你能通过这篇文章学到更多知识。更多详情敬请关注行业资讯频道。