1、概述
一个OushuDB集群管理着多个数据库(database),每个数据库又包含多个模式(schema), 一个模式包含多个对象(表,视图,函数等),所以这些对象之间的层级结构为:
database -> schema -> (tables, functions, views)
每个模式,表,视图,函数等只属于一个database。本章主要介绍每一种对象的常见用法。具体使用语法可以参见参考手册。
2、数据库
OushuDB在初始化完成后,会默认生成三个数据库,可以使用l命令查看,或者查看pg_database系统表。
1 2 3 4 5 6 7 8 |
postgres=# \l List of databases Name | Owner | Encoding | Access privileges -----------+----------+----------+------------------- postgres | ChangLei | UTF8 | template0 | ChangLei | UTF8 | template1 | ChangLei | UTF8 | (4 rows) |
其中template0和template1为模版数据库。template1为系统默认用来创建新数据库的模版数据库,用户可以修改。template0默认不接受连接,所以不可更改,目的是始终保存一个干净的模版数据库。
创建一个数据库的时候,可以指定一个数据库的模版数据库。缺省为template1,现在OushuDB只支持以template0,template1和postgres数据库为模版数据库。例如:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 |
postgres=# create database tdb; # 创建一个新数据库,默认以template0为模版 CREATE DATABASE postgres=#\c postgres # 连接postgres postgres=# create table test(i int); # 在postgres数据库中创建表test CREATE TABLE postgres=# create table test_orc(i int) with (appendonly=true, orientation=orc); # 在postgres数据库中创建ORC格式表 CREATE TABLE postgres=# create database dbnew template postgres; CREATE DATABASE postgres=#\c dbnew # 连接dbnew |
可以看到,dbnew中现在包含test表
1 2 3 4 5 6 |
dbnew=#\d List of relations Schema | Name | Type | Owner | Storage --------+------+-------+----------+------------- public | test | table | ChangLei | append only (1 row) |
3、模式
一个数据库包含多个模式(schema),而一个模式可以包含多种命名对象,比如表,数据类型,函数,操作符等。同一个对象名字可以用在不同的模式中而不产生冲突。比如schema1中可以包含表test,schema2中也可以同时包含名字为test的表。从这个意义上,模式很像一个命名空间(namespace)。
当创建一个对象时,默认被放置在public模式中。下面是系统默认创建的schema。
1 2 3 4 5 6 7 8 9 10 11 12 |
template1=# \dn List of schemas Name | Owner --------------------+---------- hawq_toolkit | ChangLei information_schema | ChangLei pg_aoseg | ChangLei pg_bitmapindex | ChangLei pg_catalog | ChangLei pg_toast | ChangLei public | ChangLei (7 rows) |
通常在这样几个场景下,用户需要使用模式:
-
允许多个用户同时使用一个数据库,而不产生名字冲突。
-
把数据库对象组织成多个schema,好像是多个命名空间一样
-
第三方应用可以把它们的对象放到一个单独的schema中,而不和其他对象产生从图。
注意:schema不可以嵌套,也就是说,schema中不可以再包含schema。
下面是创建schema的例子。
1 |
create schema myschema; |
创建或者存取一个schema中的对象,可以使用{schema}.{object}形式,例如:
1 2 |
create table myschema.test(i int); select * from myschema.test; |
删除一个空的schema,可以使用:
1 |
drop schema myschame; |
删除不空的schema,可以使用cascade关键词:
1 |
drop schema myschema cascade; |
使用{schema}.{object}形式,通常用起来不是很方便。可以通过设置schema搜索路径来简化。”SHOW search_path”命令可以给出当前的schema搜索路径。”SET search_path TO schema-name1, schema-name2”可以设置schema搜索路径。例如:
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 |
postgres=# show search_path; search_path ---------------- "$user",public (1 row) postgres=# create schema myschema; CREATE SCHEMA postgres=# set search_path = public, myschema; SET postgres=# show search_path; search_path ------------------ public, myschema (1 row) |