PostgreSQL/Greenplum数据库的物化视图

一、物化视图（MATERIALIZED VIEW）

1、PG的版本

PostgreSQL 9.3 版本之后开始支持物化视图。

2、与普通视图的区别

View 视图：虚拟，不存在实际的数据，在查询视图的时候其实是对视图内的表进行查询操作。

物化视图：实际存在，将数据存成一张表，查询的时候对这个表进行操作。物化视图内的数据需要和表的数据进行同步，这就是refresh。

二、实际操作

1、实验环境：

数据库版本：PostgreSQL 9.6.17 及 Greenplum 6.10.1

2、初始化数据

创建表，并插入数据：

tangyibo=# create table t1 (id int ,col1 varchar(10),col2 varchar(10));
CREATE TABLE
tangyibo=# create table t2 (id int ,col3 varchar(10), col4 varchar(10), col5 varchar(10));
CREATE TABLE
tangyibo=# insert into t1 values (1,'a','b'),(2,'a','b'),(3,'a','b'),(4,'a','b'),(5,'a','b');
INSERT 0 5
tangyibo=# insert into t2 values (1,'c','d','e'),(2,'c','d','e'),(3,'c','d','e'),(4,'c','d','e'),(5,'c','d','e');
INSERT 0 5
tangyibo=# select * from t1;
 id | col1 | col2
----+------+------
  1 | a    | b
  2 | a    | b
  3 | a    | b
  4 | a    | b
  5 | a    | b
(5 rows)

tangyibo=# select * from t2;
 id | col3 | col4 | col5
----+------+------+------
  1 | c    | d    | e
  2 | c    | d    | e
  3 | c    | d    | e
  4 | c    | d    | e
  5 | c    | d    | e
(5 rows)

tangyibo=#

3、创建物化视图

tangyibo=# CREATE MATERIALIZED VIEW IF NOT EXISTS mv_t1_t2 (t1_id,t2_id, col1,col2,col3,col4,col5) AS SELECT t1.id, t2.id, t1.col1,t1.col2,t2.col3,t2.col4,t2.col5 from t1,t2 where t1.id = t2.id WITH DATA;
SELECT 5
tangyibo=# select * from mv_t1_t2;
 t1_id | t2_id | col1 | col2 | col3 | col4 | col5
-------+-------+------+------+------+------+------
     1 |     1 | a    | b    | c    | d    | e
     2 |     2 | a    | b    | c    | d    | e
     3 |     3 | a    | b    | c    | d    | e
     4 |     4 | a    | b    | c    | d    | e
     5 |     5 | a    | b    | c    | d    | e
(5 rows)

注：在Greenplum中不支持IF NOT EXISTS；

4、刷新物化视图

tangyibo=# insert into t1 values (11,'x','y');
INSERT 0 1
tangyibo=# insert into t2 values (11,'x','y','z');
INSERT 0 1
tangyibo=#

对表进行操作，不改变物化视图中的数据。查询物化视图，数据没有改变:

tangyibo=# select * from mv_t1_t2 ;
 t1_id | t2_id | col1 | col2 | col3 | col4 | col5
-------+-------+------+------+------+------+------
     1 |     1 | a    | b    | c    | d    | e
     2 |     2 | a    | b    | c    | d    | e
     3 |     3 | a    | b    | c    | d    | e
     4 |     4 | a    | b    | c    | d    | e
     5 |     5 | a    | b    | c    | d    | e
(5 rows)

tangyibo=#

(1) 全量更新：

刷新物化视图才能使物化视图的数据改变。

tangyibo=# REFRESH MATERIALIZED VIEW mv_t1_t2 WITH DATA;
REFRESH MATERIALIZED VIEW
tangyibo=# select * from mv_t1_t2 ;
 t1_id | t2_id | col1 | col2 | col3 | col4 | col5
-------+-------+------+------+------+------+------
     1 |     1 | a    | b    | c    | d    | e
     2 |     2 | a    | b    | c    | d    | e
     3 |     3 | a    | b    | c    | d    | e
     4 |     4 | a    | b    | c    | d    | e
     5 |     5 | a    | b    | c    | d    | e
    11 |    11 | x    | y    | x    | y    | z
(6 rows)

tangyibo=#

(2) 增量更新

在PostgreSQL数据库只有当物化视图中存在unique index的时候，refresh物化视图才能使用增量更新，加入concurrently参数。否则报错。

tangyibo=# REFRESH MATERIALIZED VIEW CONCURRENTLY mv_t1_t2 WITH DATA;
ERROR:  cannot refresh materialized view "public.mv_t1_t2" concurrently
HINT:  Create a unique index with no WHERE clause on one or more columns of the materialized view.
tangyibo=# create unique index uidx_mv_id on mv_t1_t2 (t1_id );
CREATE INDEX
tangyibo=# insert into t1 values (12,'xx','yy');
INSERT 0 1
tangyibo=# insert into t2 values (12,'xx','yy','zz');
INSERT 0 1
tangyibo=# REFRESH MATERIALIZED VIEW CONCURRENTLY mv_t1_t2 WITH DATA;
REFRESH MATERIALIZED VIEW
tangyibo=# select * from mv_t1_t2 ;
 t1_id | t2_id | col1 | col2 | col3 | col4 | col5
-------+-------+------+------+------+------+------
     1 |     1 | a    | b    | c    | d    | e
     2 |     2 | a    | b    | c    | d    | e
     3 |     3 | a    | b    | c    | d    | e
     4 |     4 | a    | b    | c    | d    | e
     5 |     5 | a    | b    | c    | d    | e
    11 |    11 | x    | y    | x    | y    | z
    12 |    12 | xx   | yy   | xx   | yy   | zz
(7 rows)

tangyibo=#

在Greenplum数据库中，需要唯一索引与分布式键兼容，否在会报：UNIQUE and DISTRIBUTED RANDOMLY are incompatible

虽然在PostgreSQL数据库中可以用WITH NO DATA刷新物化视图，但是在查询时会报错：

tangyibo=# REFRESH MATERIALIZED VIEW mv_t1_t2 WITH NO DATA;
REFRESH MATERIALIZED VIEW
tangyibo=# select * from mv_t1_t2 ;
ERROR:  materialized view "mv_t1_t2" has not been populated
HINT:  Use the REFRESH MATERIALIZED VIEW command.
tangyibo=# REFRESH MATERIALIZED VIEW mv_t1_t2 WITH DATA;
REFRESH MATERIALIZED VIEW
tangyibo=# select * from mv_t1_t2 ;
 t1_id | t2_id | col1 | col2 | col3 | col4 | col5
-------+-------+------+------+------+------+------
     1 |     1 | a    | b    | c    | d    | e
     2 |     2 | a    | b    | c    | d    | e
     3 |     3 | a    | b    | c    | d    | e
     4 |     4 | a    | b    | c    | d    | e
     5 |     5 | a    | b    | c    | d    | e
    11 |    11 | x    | y    | x    | y    | z
    12 |    12 | xx   | yy   | xx   | yy   | zz
(7 rows)

tangyibo=#