应用拥堵，请刘大指点2个ODA 的AWR，谢谢

whutabs

偶有一台ODA V1，最近出现问题，前台业务经常拥堵或者中断，下面是2个相关时间段的AWR

请各位大侠指点迷津，谢谢。

whutabs

AWR如附件

whutabs

AWR如附件。

201403awr.rar (104.61 KB, 下载次数: 1980)

2014-3-24 15:34 上传

点击文件名下载附件

awrrpt_1_14698_14699.html (630.07 KB, 下载次数: 515)

2014-3-24 15:33 上传

点击文件名下载附件

sheep_yang678

select account, sno, bankacc, identityid, cardbalance, ebagamt from account@dl_syntong where CardID=2729662019
这类SQL太多，通过DBLINK访问数据还没绑定变量。

Benson2014

Your database workload was not high at all.  

The database performance may be caused by network performance.

Please check the below SQL plan, whether  it is full table scan. Whether there is an index in CardID column. Over 1 min to return the result. Is it that normal?
select account, sno, bankacc, identityid, cardbalance, ebagamt from account@dl_syntong where CardID=2729662019

please check the two databases servers network by scp or ftp.

Benson2014

RAC cache fusion performance was bad,  below RAC wait was high, please check the network between two RAC.
Avg message sent queue time (ms):
Avg message sent queue time on ksxp (ms):
Avg message received queue time (ms):

% of indirect sent messages is up to 99%,  it is abnormal, any application change or database change?

could you provide the some AWR report before ago when the database performance was normal?

whutabs

非常感谢楼上二位的解答。

我个人分析，14623负载很轻，系统比较正常。从TOP5来看，DBLINK及网络有很大嫌疑。
软解析不高，存在行锁问题，网络和SQL作业有很大嫌疑。

从14698来看，系统负载较高，从TOP5来看，DBLINKE和网络也有很大嫌疑。
存在大量硬解析和失败解析，存在游标争用问题，估计是大量硬解析导致的。
从SQL上可以看到大量相同语句。

这个机器是ORACLE 的ODA一体机，内置存储，万兆内联，固态硬盘日志，应该不会存在内联问题。

请各位批评指正。