揭秘备份与恢复的原理.docx

其实一句话就可以说明白：那就是数据文件的头上不仅包含了checkpoint_change#,更重要的是它包含了这个checkpoint_change#所在的logfile的sequence＃，准确的说是rba。有了rba，在恢复时就能准确的知道到底需要哪个logfile（archivelog or redo）。结果花了很大篇幅，只想以试验的方式做个简单的验证，便于大家理解。欢迎拍砖！另外提个问题：是否存在一些数据字典它是源于redo的？--controlfile中记录的checkpoint_change#SQL select checkpoint_change# from v$database;CHECKPOINT_CHANGE#------------------?1951985--datafile中记录的checkpoint_change#SQL select checkpoint_change#,checkpoint_time from v$datafile_header;CHECKPOINT_CHANGE# CHECKPOINT_TIME------------------ -------------------?1951985 2008/10/01 14:17:00?1951985 2008/10/01 14:17:00?1951985 2008/10/01 14:17:00?1951985 2008/10/01 14:17:00--controlfile中记录的每一个datafile的checkpoint_change#SQL select checkpoint_change#,checkpoint_time from v$datafile;CHECKPOINT_CHANGE# CHECKPOINT_TIME------------------ -------------------?1951985 2008/10/01 14:17:00?1951985 2008/10/01 14:17:00?1951985 2008/10/01 14:17:00?1951985 2008/10/01 14:17:00--controlfile中记录的redo的checkpoint_change#SQL select checkpoint_change#,checkpoint_time from v$thread;CHECKPOINT_CHANGE# CHECKPOINT_TIME------------------ -------------------?1951985 2008/10/01 14:17:00SQL alter system checkpoint;系统已更改。--检查点发生之后，上面提到的checkpoint_change#都给更新SQL select checkpoint_change# from v$database;CHECKPOINT_CHANGE#------------------?1955601SQL select checkpoint_change#,checkpoint_time from v$datafile_header;CHECKPOINT_CHANGE# CHECKPOINT_TIME------------------ -------------------?1955601 2008/10/01 16:15:26?1955601 2008/10/01 16:15:26?1955601 2008/10/01 16:15:26?1955601 2008/10/01 16:15:26SQL select checkpoint_change#,checkpoint_time from v$datafile;CHECKPOINT_CHANGE# CHECKPOINT_TIME------------------ -------------------?1955601 2008/10/01 16:15:26?1955601 2008/10/01 16:15:26?1955601 2008/10/01 16:15:26?1955601 2008/10/01 16:15:26SQL select checkpoint_change#,checkpoint_time from v$thread;CHECKPOINT_CHANGE# CHECKPOINT_TIME------------------ -------------------?1955601 2008/10/01 16:15:26SQL shutdown immediate数据库已经关闭。已经卸载数据库。/pubtree/?node=1ORA