诊断一次Oracle日志切换频繁的问题

    日志切换，就是生成的日志太大，数据块的变化太频繁。

	Snap Id	Snap Time	Sessions	Cursors/Session
Begin Snap:	14560	09-Dec-15 04:00:48	59	4.5
End Snap:	14561	09-Dec-15 05:00:59	54	4.6
Elapsed:		60.19 (mins)
DB Time:		82.47 (mins)

   

  1s产生2M的日志。

	Per Second	Per Transaction	Per Exec	Per Call
DB Time(s):	1.4	3.3	0.02	0.47
DB CPU(s):	1.1	2.5	0.01	0.36
Redo size (bytes):	2,315,310.3	5,548,471.3

一个小时日志切换31次。

Statistic	Total	per Hour
log switches (derived)	31	30.90

重点来了，看segments by DB Blocks changes,看是什么对象造成的。

Segments by DB Blocks Changes

• % of Capture shows % of DB Block Changes for each top segment compared
• with total DB Block Changes for all segments captured by the Snapshot 

Owner	Tablespace Name	Object Name	Subobject Name	Obj. Type	DB Block Changes	% of Capture
TC	TBS_TCI	VIEW_TICKET		TABLE	7,758,832	36.44
TC	TBS_TCI	V_DATA_RANGE		TABLE	7,076,112	33.23
TC	TBS_TCI	MV_TCM_WORKFORM		TABLE	3,593,664	16.88
TBM	TBS_TBM	VIEW_DATAAUTH_ROLE_USER		TABLE	2,368,016	11.12
TC	TBS_TC	RPT_ACCRUALS_DETAIL		TABLE	87,648	0.41

通过这些对象找到相应的SQL语句，定位就是它们。对这种业务，有两种方法，1. delete改为truncate   2.加大redo log file的大小
36p9tcbryxnq4 delete from "VIEW_TICKET" 
213rsbk7216pc delete from "V_DATA_RANGE" 
0mv6h8360myzu delete from "MV_TCM_WORKFORM" 
2u4qz8gfqs1z7 INSERT /*+ BYPASS_RECURSIVE_CHECK */ INTO "MV_TCM_WORKFORM" select t.*, sysdate as s from v_tcm_workform t