对于DML操作来说，索引对于数据库是一个性能负担.如果索引没有被有效的使用,那么其存在性就值得从新考虑.
1. 从Oracle9i开始,Oracle允许你监视索引的使用: SQL> connect scott/tiger@connerConnected to Oracle9i Enterprise Edition Release 9.2.0.4.0 Connected as scottSQL> select index_name from user_indexes;INDEX_NAME------------------------------PK_DEPTPK_EMP开始监视pk_dept索引:SQL> alter index pk_dept monitoring usage;Index altered在此过程中,如果查询使用索引,将会记录下来:SQL> select * from dept where deptno=10;DEPTNO DNAME LOC------ -------------- ------------- 10 ACCOUNTING NEW YORK停止监视:SQL> alter index pk_dept nomonitoring usage;Index altered查询索引使用情况,YES表示在监视过程中索引被使用到:SQL> select * from v$object_usage;INDEX_NAME TABLE_NAME MONITORING USED START_MONITORING END_MONITORING----------------- ------------------ ---------- ---- ------------------- -------------------PK_DEPT DEPT NO YES 10/28/2004 10:55:19 10/28/2004 10:55:47SQL>
2.Oracle9i的Bug 在9205之前，如果你不慎监控了SYS.I_OBJAUTH1索引，并且不幸在重起数据库之前没有停止它，那么你的数据库将会无法启动，并且不会给出任何错误信息。
以下这条简单的语句可以轻易再现这个问题:
'ALTER INDEX SYS.I_OBJAUTH1 MONITORING USAGE'
如果你有了足够好的备份(严重警告，请不要拿你的生产数据库进行测试)，你可以尝试一下:
[oracle@jumper oradata]$ sqlplus "/ as sysdba"
SQL*Plus: Release 9.2.0.4.0 - Production on Sat Dec 4 10:09:30 2004
Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved.
Connected to:Oracle9i Enterprise Edition Release 9.2.0.4.0 - ProductionWith the Partitioning optionJServer Release 9.2.0.4.0 - Production
SQL> alter index SYS.I_OBJAUTH1 monitoring usage ;
Index altered.
SQL> shutdown immediate;Database closed.Database dismounted.ORACLE instance shut down.SQL> startupORACLE instance started.
Total System Global Area 80811208 bytesFixed Size 451784 bytesVariable Size 37748736 bytesDatabase Buffers 41943040 bytesRedo Buffers 667648 bytesDatabase mounted.
此时，数据库挂起，而且不会有任何提示，在alert<sid>.log文件中，你可以看到:
[oracle@jumper bdump]$ tail -f alert_conner.log Completed: ALTER DATABASE MOUNTSat Dec 4 10:09:49 2004ALTER DATABASE OPENSat Dec 4 10:09:49 2004LGWR: Primary database is in CLUSTER CONSISTENT modeThread 1 opened at log sequence 54 Current log# 2 seq# 54 mem# 0: /opt/oracle/oradata/conner/redo02.logSuccessful open of redo thread 1.Sat Dec 4 10:09:49 2004SMON: enabling cache recoverySat Dec 4 10:10:33 2004Restarting dead background process QMN0QMN0 started with pid=9
然后数据库将会停在此处。
如果不知道此bug存在，你可能会一筹莫展的。
现在你能做的就是从备份中恢复，或者升级到9.2.0.5。
Oracle已经Release了这个Bug，你可以参考Metalink:Note:2934068.8,Oracle声明在9.2.0.5 (Server Patch Set)和 10g Production Base Release中fixed了这个Bug。
[oracle@jumper oradata]$ rm -rf conner[oracle@jumper oradata]$ cp -R connerbak/ conner[oracle@jumper oradata]$ sqlplus '/ as sysdba'
SQL*Plus: Release 9.2.0.4.0 - Production on Sat Dec 4 10:19:07 2004
Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved.
Connected to an idle instance.
SQL> startupORACLE instance started.
Total System Global Area 80811208 bytesFixed Size 451784 bytesVariable Size 37748736 bytesDatabase Buffers 41943040 bytesRedo Buffers 667648 bytesDatabase mounted.Database opened.SQL>
3. 在特殊的情况下,你可能需要清除这个v$object_usage视图中的信息. Oracle的说法是,在下一次收集该对象的索引使用情况时会自动覆盖上一次的信息,不提供清除手段.
稍微研究了一下.
v$object_usage是基于以下基表建立起来的:
create or replace view v$object_usage(index_name, table_name, monitoring, used, start_monitoring, end_monitoring)asselect io.name, t.name, decode(bitand(i.flags, 65536), 0, 'NO', 'YES'), decode(bitand(ou.flags, 1), 0, 'NO', 'YES'), ou.start_monitoring, ou.end_monitoringfrom sys.obj$ io, sys.obj$ t, sys.ind$ i, sys.object_usage ouwhere io.owner# = userenv('SCHEMAID') and i.obj# = ou.obj# and io.obj# = ou.obj# and t.obj# = i.bo#/
注意到v$object_usage关键信息来源于OBJECT_USAGE表.另外我们可以注意一下,此处v$object_usage的查询基于userenv('SCHEMAID')建立.所以以不同用户登录,你是无法看到其他用户的索引监视信息的,即使是dba,但是可以从object_usage表中得到.
SQL> select * from v$object_usage;INDEX_NAME TABLE_NAME MON USE START_MONITORING END_MONITORING------------------------------ ------------------------------ --- --- ------------------- -------------------PK_DEPT DEPT NO YES 10/28/2004 10:55:19 10/28/2004 10:55:47SQL> select * from object_usage;select * from object_usage *ERROR at line 1:ORA-00942: table or view does not existSQL> connect /as sysdbaConnected.SQL> / OBJ# FLAGS START_MONITORING END_MONITORING---------- ---------- ------------------- ------------------- 6288 1 10/28/2004 10:55:19 10/28/2004 10:55:47
实际上我们清除了object_usage表的记录,实际上也就清空了v$object_usage的信息.
SQL> delete from object_usage;1 row deleted.SQL> commit;Commit complete.SQL> select * from v$object_usage;no rows selected
此操作对数据库没有潜在的影响,但是请谨慎使用.作为实验目的提供.
