数据库上云实践：使用Ora2pg进行数据库迁移

本文将介绍如何使用 Oracle到postgresql的迁移使用ora2pg工具。配置环境后，你将了解如何编写配置文件，进行数据库迁移任务。

重要

• 在本教程中，你将了解如何执行以下操作：
• 编写ora2pg配置文件
• 通过配置文件将Oracle中的表、序列、视图、同义词、函数过程包进行迁移

前置条件

• perl解释器
• perl 模块 DBD(DataBase Description) *DBI(DataBase Interface)
• ora2pg源码

配置环境

Win环境配置

1. perl解释器安装: strawberry-perl-5.32.1.1-64bit.msi
2. 在win下 strawberry-perl(version 5.32以后)自带DBD::Oracle 和DBD::Pg.

perl -MCPAN -e install DBD::Oracle
perl -MCPAN -e install DBD::Pg

3. 利用perl解释器，解释makefile.pl文件，配置环境并安装ora2pg。win下使用gmake指令

cd ora2pg 
perl Makefile.PL
gmake && gmake install

4. DBI的安装需要下载源码包，通过源码安装(在win下，可以使用gmake指令)

tar -zxvf DBI-1.641.tar.gz
cd DBI-1.641
perl Makefile.PL
gmake && gmake install

5. 安装Oracle客户端

linux环境配置

1. perl版本5.10以上，安装依赖包perl-ExtUtils-CBuilderperl-ExtUtils-MakeMaker(注：win下的strawberry-perl自带这两个包)

   yum install -y perl perl-ExtUtils-CBuilder perl-ExtUtils-MakeMaker

2. 安装DBI

   tar -zxvf DBI-1.641.tar.gz
   cd DBI-1.641
   perl Makefile.PL
   make && make install

3. 安装Oracle客户端(略)

       rpm -ivh oracle-instantclient12.2-basic-12.2.0.1.0-1.x86_64.rpm
       rpm -ivh oracle-instantclient12.2-devel-12.2.0.1.0-1.x86_64.rpm
       rpm -ivh oracle-instantclient12.2-jdbc-12.2.0.1.0-1.x86_64.rpm
       rpm -ivh oracle-instantclient12.2-sqlplus-12.2.0.1.0-1.x86_64.rpm

4. 安装DBD-Oracle

   定义ORACLE_HOME环境变量

   tar -zxvf DBD-Oracle-1.74.tar.gz
   cd DBD-Oracle-1.74
   perl Makefile.PL
   make && make install

5. 安装DBD:PG

   #使用官方提供的yum安装方式
   #Linux系统版本centos7 64位版本
   #Pg版本为12
   yum install -y https://download.postgresql.org/pub/repos/yum/reporpms/EL-7-x86_64/pgdg-redhat-repo-latest.noarch.rpm
   yum install postgresql12
   yum install postgresql12-server
   yum install postgresql12-contrib
   
   #配置开机启动与启动
   systemctl enable postgresql-12
   systemctl start postgresql-12
   #修改配置文件
   #打开并编辑文件“/var/lib/pgsql/12/data/postgresql.conf”。将“#listen_addresses = ‘localhost’”改为“listen_addresses = ‘*’”
   #打开并编辑文件“/var/lib/pgsql/12/data/pg_hba.conf”。在文件的末尾添加“host all all 0.0.0.0/0 md5”
   #设置pgsql的环境变量

6. 安装Ora2PG

   tar -zxvf ora2pg-22.1.tar.bz2cd ora2pg-22.1/perl Makefile.PLmake && make install

Ora2Pg使用方法

编写conf文件，执行指令将Oracle转化为Postgresql

ora2pg -c ora2pg_table.conf

配置文件ora2pg_table.conf

PG_VERSION 12
ORACLE_HOME  /home/oracle/app/oracle/product/11.2.0/dbhome_1
ORACLE_DSN  dbi:Oracle:host=xx.xx.x.xxx;service_name=xxx;port=xx
ORACLE_USER xxxxxx
ORACLE_PWD  ######
USER_GRANTS 1
SCHEMA  xxxx
TYPE TABLE
PG_NUMERIC_TYPE    0
PG_INTEGER_TYPE    1
DEFAULT_NUMERIC float
#SKIP    fkeys pkeys ukeys indexes checks
NLS_LANG    AMERICAN_AMERICA.UTF8
OUTPUT    /home/ora2pg/xxxxx.sql

配置文件解释：Oracle数据库连接配置项

ORACLE_HOME

使用ORACLE_HOME作为环境变量，ora2pg可以通过该参数找到Perl的DBD模块连接ORACLE数据库所需要的依赖包。

• 若在Linux环境下，该变量可能为： /home/oracle/app/oracle/product/11.2.0/dbhome_1
• 若在Windows环境下，该变量可能为：D:\app\Administrator\product\11.2.0\dbhome_1

ORACLE_DSN

该参数与Oracle数据库具体配置有关，主要为数据库的

host,sid,port等参数

dbi:Oracle:host=oradb_host.myhost.com;sid=DB_SID;port=1521dbi:Oracle:DB_SID
#On 18c this could be for example:dbi:Oracle:host=192.168.1.29;service_name=pdb1;port=1521
#for the second notation the SID should be declared in the well known
#file $ORACLE_HOME/network/admin/tnsnames.ora or in the path given to#the TNS_ADMIN environment variable.
#For MySQL the DSN will lool like this:dbi:mysql:host=192.168.1.10;database=sakila;port=3306
the 'sid' part is replaced by 'database'.

ORACLE_USER et ORACLE_PWD

USER和PWD表示Oracle数据库的用户名和密码，选用账号的时候尽量选择DBA权限的账号，以防止数据库扫描时出现的权限问题。

• 若配置文件*.conf未设置该参数，可安装perl的Term::ReadKey模块，在执行指令时的交互式命令行输入账号密码
• 连接本地的数据据库实例如果为"as sysdba"，那么可以将ORACLE_USER设置为""，密码为空

USER_GRANTS

• 如果连接Oracle数据库为普通，该值设置为1，此时没有权限抽取DBA 表格，此时将会使用ALL 表格
• 若导出类型为GRANT，该值必须设置为0，否者将无效。

TRANSACTION*

事务隔离类型，如果需要改变导出数据的默认隔离级别，需要设置该参数，总共由4中事务隔离类型：

• READ ONLY
• READ WRITE
• ISOLATION LEVEL SERIALIZABLE
• ISOLATION LEVEL READ COMMITTED

基本原理

在Oarcle数据库和Postgresql数据库中间，使用一台机器，这台机器实现将Oracle数据库的元数据导出，并根据数据库中的表格、序列、视图等内容生成符合Postgresql语法的DDL(data definition language)，并保存为相应的 *.sql文件按

数据类型变化(参考Ora2PG源码定义)

our %TYPE = (	'NUMBER' => 'numeric',
# CHAR types limit of 2000 bytes with defaults to 1 if no length
# is specified. PG char type has max length set to 8104 so it
# should match all needs
				'CHAR' => 'char',
				'NCHAR' => 'char',
# VARCHAR types the limit is 2000 bytes in Oracle 7 and 4000 in
# Oracle 8. PG varchar type has max length iset to 8104 so it
# should match all needs
				'VARCHAR' => 'varchar',
                'NVARCHAR' => 'varchar',
                'VARCHAR2' => 'varchar',
                'NVARCHAR2' => 'varchar',
                'STRING' => 'varchar',
# The DATE data type is used to store the date and time
# information. PG type timestamp should match all needs.
				'DATE' => 'timestamp',
# Type LONG is like VARCHAR2 but with up to 2Gb. PG type text
# should match all needs or if you want you could use blob
 				'LONG' => 'text',
# Character data of variable length
				'LONG RAW' => 'bytea',
# Types LOB and FILE are like LONG but with up to 4Gb. PG type
# text should match all needs or if you want you could use blob
# (large object)
				'CLOB' => 'text',
# A large object containing single-byte characters
				'NCLOB' => 'text',
# A large object containing national character set data
				'BLOB' => 'bytea',
# Binary large object
# The full path to the external file is returned if destination type is text.
# If the destination type is bytea the content of the external file is returned.
				'BFILE' => 'bytea',
# Locator for external large binary file
# The RAW type is presented as hexadecimal characters. The
# contents are treated as binary data. Limit of 2000 bytes
# PG type text should match all needs or if you want you could
# use blob (large object)、
				'RAW' => 'bytea',
                'ROWID' => 'oid',
                'UROWID' => 'oid',
                'FLOAT' => 'double precision',
                'DEC' => 'decimal',
                'DECIMAL' => 'decimal',
                'DOUBLE PRECISION' => 'double precision',
                'INT' => 'numeric',
                'INTEGER' => 'numeric',
                'BINARY_INTEGER' => 'integer',
                'PLS_INTEGER' => 'integer',
                'REAL' => 'real',
                'SMALLINT' => 'smallint',
                'BINARY_FLOAT' => 'double precision',
                'BINARY_DOUBLE' => 'double precision',
                'TIMESTAMP' => 'timestamp',
                'BOOLEAN' => 'boolean',
                'INTERVAL' => 'interval',
                'XMLTYPE' => 'xml',
                'TIMESTAMP WITH TIME ZONE' => 'timestamp with time zone',
                'TIMESTAMP WITH LOCAL TIME ZONE' => 'timestamp with time zone',
                'SDO_GEOMETRY' => 'geometry'
                );

常见的Oracle对象到Postgresql数据库的转换

ORACLE	POSTGRESQL
Database Link	Foreign Data Wrapper
External Table	Foreign Table
Synonym	View / Set search_path
Global Temporary Table	Unlogged Table / Temp Table
Virtual column	View / Function / Trigger
Connect by	With Recursive
Reverse Index	Functional Index
Index Organized Table (IOT table)	Cluster the table according to an Index

数据库链接(Database Link)

Oracle

CREATE PUBLIC DATABASE LINK remote_service USING 'remote_db';
SELECT * FROM employees@remote_service;

Postgresql

CREATE SERVER remote_service FOREIGN DATA WRAPPER oracle_fdw OPTIONS (dbserver 'remote_db');
CREATE USER MAPPING FOR current_user SERVER remote_service OPTIONS (user 'scott', password 'tiger');
CREATE FOREIGN TABLE employees_fdw (<columns_list>) SERVER remote_service OPTIONS(schema 'HR',table 'EMPLOYEES');

外部表(External Tables)

假设有这么一张表

cat /data/ext/file_ext.csv
1234,ALBERT,GRANT,21
1235,ALFRED,BLUEOS,26
1236,BERNY,JOLYSE,34

Oracle

CREATE OR REPLACE DIRECTORY ext_dir AS '/data/ext/';
CREATE TABLE ext_table (
    empno VARCHAR2(4), 
    firstname VARCHAR2(20), 
    lastname VARCHAR2(20),
    age VARCHAR2(2)
) ORGANIZATION EXTERNAL (DEFAULT DIRECTORY ext_dir ACCESS PARAMETERS (…LOCATION ('file_ext.csv')));

Postgresql

CREATE FOREIGN TABLE ext_table(
    empno VARCHAR(4),
    firstname VARCHAR(20),
    lastname VARCHAR(20),
    age VARCHAR(2)
)SERVER ext_dir OPTIONS (filename '/data/ext/file_ext.csv', format 'csv', delimiter ',');

同义词(Synonyms)

Synonyms are not supported in PostgreSQL

Oracle

CREATE PUBLIC SYNONYM emp_table FOR hr.employees [@ dblink];

Postgresql

有两种方式search path和view

--search path – session level – no permanent effect, it needs to be set for every connectionSET search_path TO other_schema;
--search path – role or database level – it takes permanent effect--@postgresqlalter database <database_name> set search_path = "other_schema";--@database_namealter role <role_name> set search_path = "other_schema";--view:CREATE VIEW public.emp_table AS SELECT * FROM hr.employees;ALTER VIEW public.emp_table OWNER TO hr;GRANT ALL ON public.emp_table TO PUBLIC;

全局临时表(Global Temporary Tables)

Option 1: Unlogged Table:

Oracle

CREATE GLOBAL TEMPORARY TABLE MY_CONTRACT_MONTH(    
    ID NUMBER(10),
    CMONTH DATE
)ON COMMIT DELETE ROWS;

PostgreSQL

CREATE UNLOGGED TABLE MY_CONTRACT_MONTH(
    ID number,
    CMONTH timestamp,
    pid bigint default pg_backend_pid()
);

ALTER TABLE MY_CONTRACT_MONTH ENABLE ROW LEVEL SECURITY;
ALTER TABLE MY_CONTRACT_MONTH FORCE ROW LEVEL SECURITY;

CREATE POLICY cm_pid ON MY_CONTRACT_MONTH TO <role_name> 
USING (pid = (selectpg_backend_pid()));

在Postgresql中Role即User，若表格不保存日志，对于 Unlogged 表，必须实现行级安全，以防止会话查询其他会话数据。还需要实施一项作业，以消除非活动会话的未记录表上的数据

DELETE FROM smdr.contract_months cm
WHERE not exists (select 1 from pg_stat_activity psa where psa.pid = cm.pid);

Option 2: Temp Table:

Oracle

CREATE GLOBAL TEMPORARY TABLE MY_CONTRACT_MONTH(
    ID NUMBER(10),
    CMONTH DATE
)ON COMMIT DELETE ROWS;
--Oracle 永久存储临时表的定义，就像常规表的定义一样。

PostgreSQL

CREATE GLOBAL TEMPORARY TABLE MY_CONTRACT_MONTH(
    ID NUMERIC,
    CMONTH TIMESTAMP
)ON COMMIT DELETE ROWS;

CREATE TEMPORARY TABLE 语句创建一个临时表，该表在会话或当前事务结束时自动删除（ON COMMIT DROP 选项。在转换过程中，您需要从应用程序代码中提取 CREATE TEMPORARY TABLE 语句，存储过程、触发器等并执行它们一次以创建临时表定义。Oracle 不支持 ON COMMIT DROP，因此如果需要此选项，则需要明确每次提交后执行 DROP TABLE 语句 ON COMMIT PRESERVE ROWS 是 PostgreSQL 中的默认值，ON COMMIT DELETE ROWS 是 Oracle 中的默认值

Virtual Column

Oracle

CREATE TABLE VIRT_COL_TABLE (
  id NUMBER,
  first_name VARCHAR2(10),
  last_name VARCHAR2(10),
  salary NUMBER(9,2),
  comm1 NUMBER(3),
  comm2 NUMBER(3),
  salary1 AS (ROUND(salary*(1+comm1/100),2)),
  salary2 NUMBER GENERATED ALWAYS AS (ROUND(salary*(1+comm2/100),2)) VIRTUAL
);

PostgreSQL

CREATE TABLE virt_col_table (
  id bigint NOT NULL,
  first_name varchar(10),
  last_name varchar(10),
  salary double precision,
  comm1 smallint,
  comm2 smallint,
  salary1 bigint,
  salary2 bigint);

Ora2pg通过一个包含触发器的定义的额外文件 VIRTUAL_COLUMNS_(...).sql，来实现原始Virtual Column的默认值

DROP TRIGGER IF EXISTS virt_col_VIRT_COL_TABLE_trigger ON VIRT_COL_TABLE CASCADE;
CREATE OR REPLACE FUNCTION fct_virt_col_VIRT_COL_TABLE_trigger() 
 RETURNS trigger
 AS $BODY$
BEGINNEW.SALARY2 = ROUND(NEW.SALARY*(1+NEW.COMM2/100),2);
NEW.SALARY1 = ROUND(NEW.SALARY*(1+NEW.COMM1/100),2);
RETURN NEW;
end 
$BODY$ LANGUAGE 'plpgsql' SECURITY DEFINER;
CREATE TRIGGER virt_col_VIRT_COL_TABLE_trigger
BEFORE INSERT OR UPDATE ON VIRT_COL_TABLE FOR EACH ROWEXECUTE PROCEDURE fct_virt_col_VIRT_COL_TABLE_trigger();

Connect By – Hierarchical query

在Oracle通过connect by (子句) 以实现层级遍历，而在PostgreSQL中使用WITH RECURSIVE (子句)

Oracle

CREATE TABLE taxonomy (
  key NUMBER(11) NOT NULL CONSTRAINT taxPkey PRIMARY KEY,
  value VARCHAR2(255),taxHier NUMBER(11)
);
ALTER TABLE taxonomy ADD CONSTRAINT taxTaxFkey FOREIGN KEY (taxHier) REFERENCEStax(key);
SELECT value FROM taxonomy CONNECT BY PRIOR key = taxHier START WITH key = 0;

PostgreSQL

WITH RECURSIVE cte AS (
  SELECT key, value, 1 AS level 
    FROM taxonomy
    WHERE key = 0
  UNION ALL
  SELECT t.key, t.value, c.level + 1
    FROM cte c 
    JOIN taxonomy t ON t.taxHier = c.key
  )
SELECT value FROM cte ORDER BY level;

反向索引(Reverse Index)

当反向索引应用于 TEXT 列时，此解决方法有效。

Oracle

CREATE TABLE REV_TEMP (
  Id NUMBER(10) NOT NULL PRIMARY KEY,
  Description VARCHAR2(512) NOT NULL
);
CREATE INDEX REV_TEMP_N1 ON REV_TEMP(Description) REVERSE;

PostgreSQL

CREATE TABLE REV_TEMP (
  Id NUMERIC(10) NOT NULL PRIMARY KEY,
  Description VARCHAR(512) NOT NULL
);
CREATE INDEX REV_TEMP_N1 ON REV_TEMP(REVERSE(Description));

索引组织表(Index Organized Table)

Oracle 数据库默认使用堆表。 可以使用 ORGANIZATION INDEX 子句创建索引组织表，Oracle 数据库始终使用主键作为集群键。

Oracle

CREATE TABLE IOT_TEMP (
  Id NUMBER(10) NOT NULL PRIMARY KEY,
  Description VARCHAR2(512) NOT NULL
) ORGANIZATION INDEX;

PostgreSQL

PostgreSQL 只使用堆表。 但是，使用 CLUSTER 子句将堆表的内容与索引对齐。

CREATE TABLE IOT_TEMP (
  Id NUMERIC(10) NOT NULL PRIMARY KEY,
  Description VARCHAR(512) NOT NULL
);
CREATE INDEX IOT_TEMP_N1 ON IOT_TEMP(ID);

job梳理与改造

SELECT JOB,WHAT,INTERVAL,SCHEMA_USER FROM ALL_JOBS WHERE SCHEMA_USER='SJCK'

Known Unsupported

• Type inheritance and type with member method are not supported
• Global indexes over partitions are not supported
• Compound triggers are not supported