1，创建数据库，全局数据库名称分别为：masterone.test.com/mastertwo.test.comSerivce name 分别为：masterone.test.com/mastertwo.test.com
2，在两个数据库中tnsnames.ora配置客户访问：配置内容如下：MASTERTWO =(DESCRIPTION =(ADDRESS = (PROTOCOL = TCP)(HOST = mastertwo.test.com)(PORT = 1521))(CONNECT_DATA =(SERVER = DEDICATED)(SERVICE_NAME = mastertwo.test.com)))MASTERONE =(DESCRIPTION =(ADDRESS = (PROTOCOL = TCP)(HOST = masterone.test.com)(PORT = 1521))(CONNECT_DATA =(SERVER = DEDICATED)(SERVICE_NAME = masterone.test.com)))3，打开SQL Plus测试链接（保证数据库可联通）>connect system@masterone;>connect system@mastertwo;
4，创建replication admin ，使用system连接masterone.test.com数据库；
5，创建repadmin 执行命令Create user repadmin identified by repadmin;
6，赋予repadmin用户可以创建和管理复制环境的权限Exec dbms_repcat_admin.grant_admin_any_schema(username=>'repadmin');Grant comment any table to repadmin;Grant lock any table to repadmin;如果使用OEM管理工具，还需执行如下命令：GRANT SELECT ANY DICTIONARY TO repadmin;
7，注册repadmin为传播函数Exec dbms_defer_sys.register_propagator(username=>'repadmin');
8，注册repadmin为接受函数BEGINDBMS_REPCAT_ADMIN.REGISTER_USER_REPGROUP (username => 'repadmin',privilege_type => 'receiver',list_of_gnames => NULL);END;
9，为了在检查时保持延迟事务队列的大小, 您应该清除已成功完成的延迟事务。SCHEDULE_PURGE 过程为您自动清除过程。使用repadmin登陆数据库Connect repadmin;BEGINDBMS_DEFER_SYS.SCHEDULE_PURGE (next_date => SYSDATE,interval => 'SYSDATE + 1/24',delay_seconds => 0);END;
10，在mastertwo.mor.cr上执行相同操作4-9；11，使用system在masterone.test.com上创建共有dblinkConnect system@masteroneCreate public datebase link mastertwo.test.com using 'mastertwo';Connect repadmin@masteroneCreate database link mastertwo.test.com connect to repadmin identified by repadmin;使用system在mastertwo.test.com上创建共有dblinkConnect system@mastertwo;Create public database link masterone.test.com using 'masterone';Connect repadmin@mastertwoCreate database link masterone.mor.cr connect to repadmin identified by repadmin;12，使用repadmin登陆masterone.test.com/mastertwo.test.comConnect repadmin@masteroneBEGINDBMS_DEFER_SYS.SCHEDULE_PUSH (destination => 'mastertwo.test.com',interval => 'SYSDATE + (1/144)',next_date => SYSDATE,parallelism => 1,execution_seconds => 1500,delay_seconds => 1200); /*延迟时间可以设置小一点*/END;Connect repadmin@mastertwoBEGINDBMS_DEFER_SYS.SCHEDULE_PUSH (destination => 'masterone.test.com',interval => 'SYSDATE + (1/144)',next_date => SYSDATE,parallelism => 1,execution_seconds => 1500,delay_seconds => 1200);END;13，创建master group在创建同步组之前应该确保复制环境的中的每个数据库中有相同的schema和表结构。例如在测试环境中定义的schema为hr，表结构为test只在master site上执行此操作：masterone.test.com作为master siteConnect repadmin@masteroneExec dbms_repcat.create_master_repgroup('hrgroup')
14，向master group中添加objectBEGINDBMS_REPCAT.CREATE_MASTER_REPOBJECT (gname => 'hrgroup',type => 'TABLE',oname => 'test',sname => 'hr',use_existing_object => TRUE,copy_rows => FALSE);END;15，添加额外的master siteBEGINDBMS_REPCAT.ADD_MASTER_DATABASE (gname => 'hrgroup',master => 'mastertwo.test.com',use_existing_objects => TRUE,copy_rows => FALSE,propagation_mode => 'ASYNCHRONOUS');END;使用如下命令检查，如果出现两个数据库的连接则操作成功SELECT DBLINK FROM DBA_REPSITES WHERE GNAME = 'HRGROUP';mastertwo.test.commasterone.test.com16，添加复制支持BEGIN DBMS_REPCAT.GENERATE_REPLICATION_SUPPORT (sname => 'hr',oname => 'test', type => 'TABLE',min_communication => TRUE); END;完成检查通过命令检查，返回0可继续执行SELECT COUNT(*) FROM DBA_REPCATLOG WHERE GNAME = 'HRGROUP';
17，启用复制BEGIN DBMS_REPCAT.RESUME_MASTER_ACTIVITY (gname => 'hrgroup'); END;
18，测试在masterone.test.com上hr.test表中分别通过进行增删改操作Insert into hr.test values (1,'testuser1');Commit;(在SQL Plus上操作别忘提交)由于复制延迟为1200，毫秒所以不会很快同步到mastertwo.test.com上