增量检查点的作用是为了均衡负载,由fast_start_mttr_target这个参数触发,增量渐进写出。所以,CHECKPOINT_CHANGE#会有延迟,不会马上更新。
下面用三种方法证明:
法一:
理论:日志却会有active,即:被RBA指针覆盖的日志。如是完全检查点,则RBA会一下子干到重做日志组的最后一条,没有了RBA,那么日志的状态便是inactive了;而如是增量检查点,则RBA会慢慢下移,有被RBA覆盖的都是active。Oracle总是希望RBA与重做日志组的最后一条的距离最短,增量检查点就是时不时要移动它。
实验:
SQL> alter system switch logfile;
System altered
SQL> select group#,status from v$log;
GROUP# STATUS
---------- ----------------
1 CURRENT
2 INACTIVE
3 ACTIVE
SQL> alter system switch logfile;
System altered
SQL> select group#,status from v$log;
GROUP# STATUS
---------- ----------------
1 ACTIVE
2 CURRENT
3 ACTIVE
法二:
理论:增量检查点没有全部写,所以checkpoint_change#没有马上更新;但完全检查点,因为全部写,checkpoint_change#会马上更新。
实验:
SQL> select checkpoint_change# from v$database;
CHECKPOINT_CHANGE#
------------------
1716093
SQL> alter system checkpoint;
System altered
SQL> select checkpoint_change# from v$database;
CHECKPOINT_CHANGE#
------------------
1716275
法三:
实验:
SQL> show parameter log_checkpoints_to_
NAME TYPE VALUE
------------------------------------ ----------- ------------------------------
log_checkpoints_to_alert boolean FALSE
SQL> alter system set log_checkpoints_to_alert=true;
System altered
SQL> alter system checkpoint;
System altered
手工触发一个完全检查点,告警日志记录如下:
Thu Jun 07 01:27:17 2012
Beginning global checkpoint up to RBA [0x2d.1e4.10], SCN: 1716634
Completed checkpoint up to RBA [0x2d.1e4.10], SCN: 1716634
接下来手工触发一个增量检查点:
SQL> alter system switch logfile;
System altered
告警日志如下:
Thu Jun 07 01:28:15 2012
Beginning log switch checkpoint up to RBA [0x2e.2.10], SCN: 1716660
Thread 1 advanced to log sequence 46
Current log# 3 seq# 46 mem# 0: D:ORACLEPRODUCT10.2.0ORADATAORCLONLINELOGO1_MF_3_7TQZWZOY_.LOG
Current log# 3 seq# 46 mem# 1: D:ORACLEPRODUCT10.2.0FLASH_RECOVERY_AREAORCLONLINELOGO1_MF_3_7TQZX11D_.LOG
从这两个告警日志中,我们可以看到,完全检查点会马上将RBA下移,而增量检查点就会悠着点了。
更多Oracle相关信息见 专题页面