ORACLE10G rac 两节点。在其中的一个节点在部署了流复制。这一节点上的响应时间比另一节点大很多。
下面是此节点的TOP5事件。
Top 5 Timed Events
Avg %Total
~~~~~~~~~~~~~~~~~~
wait Call
Event
WaitsTime (s) (ms) Time Wait Class
------------------------------ ------------ ----------- ------ ------ ----------
Streams capture: waiting for a2,292 1,272555 21.4Other
CPU time
1,216
20.5
log file sequential read
19,246 254 134.3 System I/O
db file sequential read
22,077 16172.7 User I/O
gcs log flush sync
40,867
7921.3Other
以下是与SQL ORDER BY READS
SQL ordered by Reads
DB/Inst: BILLBJ/billbj2Snaps: 7505-7506
-> Total Disk Reads:
24,716
-> Captured SQL account for114.3% of Total
Reads
CPU Elapsed
Physical ReadsExecutionsper Exec %Total Time (s)Time (s)SQL Id
-------------- ----------- ------------- ------ -------- --------- -------------
9,295
39 238.3 37.6 423.97523.03 2q0h3h6ptrcmx
INSERT INTO BANDWIDTHTABLE(ID,CHANNELID,BUSINESSTYPE,DAY,NODEID,DATA,INDATA,USER
ID,PARFLAG) values(BANDWIDTH_SEQUENCE.NEXTVAL,:1,:2,:3,:4,:5,:6,:7,:8)
9,235 204,146 0.0 37.446.64 70.95 ayxg8ajztf866
insert into system.logmnr_restart_ckpt$ (session#, valid, ckpt_scn, xidusn, xids
lt, xidsqn, session_num, serial_num, ckpt_info, flag, offset, client_data, spare
1) values (:1, 0, :2, :3, :4, :5, :6, :7, EMPTY_BLOB(), :8, 1, EMPTY_BLOB(), :9)
8,278 497,496 0.0 33.5 190.35237.80 gdc5f827s4u6t
MERGE INTO BANDWIDTH_DAILYTABLE A USING (SELECT :B7 CHANNELID, :B6 BUSINESSTYPE,
:B5 DAY, :B4 NODEID, :B3 DATA, :B2 INDATA, :B1 USERID, :B8 PARFLAG FROM DUAL) T
ON (A.USERID = T.USERID AND A.PARFLAG = T.PARFLAG AND A.CHANNELID = T.CHANNELID
AND A.DAY = T.DAY AND A.NODEID = T.NODEID AND A.BUSINESSTYPE = T.BUSINESSTYPE)
我感觉这个问题跟流复制有很大的关系，但不知原因，谁清楚的？
|
