Discrepancy between v$sysmetric and v$sysstat
- From: "Allen, Brandon" <Brandon.Allen@xxxxxxxxxxx>
- To: "'oracle-l@xxxxxxxxxxxxx'" <oracle-l@xxxxxxxxxxxxx>
- Date: Wed, 16 Sep 2009 11:17:42 -0700
I'm just now getting around to digging into the metrics introduced in 10g via
v$sysmetric and related views and I'm noticing a pretty big discrepancy between
the values for metric "Physical Reads Per Sec" and the values derived from
"physical reads" in v$sysstat. For example, if I look at a statspack report
for a given time period, I see the following indicating the average was 26.99
reads per second:
Snapshot Snap Id Snap Time
~~~~~~~~ ---------- ------------------
Begin Snap: 12192 11-Sep-09 14:15:02
End Snap: 12195 11-Sep-09 15:00:02
Elapsed: 45.00 (mins)
Load Profile Per Second Per Transaction
~~~~~~~~~~~~ --------------- ---------------
Physical reads: 26.99 318.17
Statistic Total per Second per Trans
--------------------------------- ------------------ -------------- ------------
physical reads 72,862 27.0 318.2
But if I look at v$sysmetric_history for the same period, I get the results
shown below and if I take the average of all those values I only get 15.9 reads
per second. I've repeated this for several different time ranges and the
numbers never match up on this system, which is running Oracle 10.2.0.4 on
Oracle Linux 5.2, but when I try the same comparison on another system running
Oracle 10.2.0.2 on AIX, the numbers match up almost perfectly. I already
searched Metalink for bugs and couldn't find anything. Has anyone else noticed
this problem? Any idea what might be causing it?
Thanks,
Brandon
SYS@d517>select begin_time, end_time, value from v$sysmetric_history where
metric_name = 'Physical Reads Per Sec' and intsize_csec > 5900 order by 1;
BEGIN_TIME END_TIME VALUE
------------------ ------------------ ----------
...
11-SEP-09 14:15:17 11-SEP-09 14:16:18 0
11-SEP-09 14:16:18 11-SEP-09 14:17:18 498.518395
11-SEP-09 14:17:18 11-SEP-09 14:18:18 .03329449
11-SEP-09 14:18:18 11-SEP-09 14:19:18 0
11-SEP-09 14:19:18 11-SEP-09 14:20:18 119.843516
11-SEP-09 14:20:18 11-SEP-09 14:21:18 0
11-SEP-09 14:21:18 11-SEP-09 14:22:18 0
11-SEP-09 14:22:18 11-SEP-09 14:23:18 .08322237
11-SEP-09 14:23:18 11-SEP-09 14:24:18 0
11-SEP-09 14:24:18 11-SEP-09 14:25:17 0
11-SEP-09 14:25:17 11-SEP-09 14:26:17 0
11-SEP-09 14:26:17 11-SEP-09 14:27:17 7.20585788
11-SEP-09 14:27:17 11-SEP-09 14:28:17 .532534532
11-SEP-09 14:28:17 11-SEP-09 14:29:18 0
11-SEP-09 14:29:18 11-SEP-09 14:30:18 0
11-SEP-09 14:30:18 11-SEP-09 14:31:18 0
11-SEP-09 14:31:18 11-SEP-09 14:32:18 .049933422
11-SEP-09 14:32:18 11-SEP-09 14:33:18 0
11-SEP-09 14:33:18 11-SEP-09 14:34:18 0
11-SEP-09 14:34:18 11-SEP-09 14:35:18 0
11-SEP-09 14:35:18 11-SEP-09 14:36:18 0
11-SEP-09 14:36:18 11-SEP-09 14:37:18 53.7527043
11-SEP-09 14:37:18 11-SEP-09 14:38:17 .084616686
11-SEP-09 14:38:17 11-SEP-09 14:39:17 2.24700399
11-SEP-09 14:39:17 11-SEP-09 14:40:17 0
11-SEP-09 14:40:17 11-SEP-09 14:41:17 0
11-SEP-09 14:41:17 11-SEP-09 14:42:18 3.04695305
11-SEP-09 14:42:18 11-SEP-09 14:43:18 0
11-SEP-09 14:43:18 11-SEP-09 14:44:18 0
11-SEP-09 14:44:18 11-SEP-09 14:45:18 0
11-SEP-09 14:45:18 11-SEP-09 14:46:18 0
11-SEP-09 14:46:18 11-SEP-09 14:47:18 .04995005
11-SEP-09 14:47:18 11-SEP-09 14:48:18 0
11-SEP-09 14:48:18 11-SEP-09 14:49:18 0
11-SEP-09 14:49:18 11-SEP-09 14:50:18 .04995005
11-SEP-09 14:50:18 11-SEP-09 14:51:18 .083250083
11-SEP-09 14:51:18 11-SEP-09 14:52:18 0
11-SEP-09 14:52:18 11-SEP-09 14:53:17 0
11-SEP-09 14:53:17 11-SEP-09 14:54:17 0
11-SEP-09 14:54:17 11-SEP-09 14:55:17 0
11-SEP-09 14:55:17 11-SEP-09 14:56:17 0
11-SEP-09 14:56:17 11-SEP-09 14:57:17 2.43090243
11-SEP-09 14:57:17 11-SEP-09 14:58:18 24.9833555
11-SEP-09 14:58:18 11-SEP-09 14:59:18 3.79683597
11-SEP-09 14:59:18 11-SEP-09 15:00:18 .116550117
...
Privileged/Confidential Information may be contained in this message or
attachments hereto. Please advise immediately if you or your employer do not
consent to Internet email for messages of this kind. Opinions, conclusions and
other information in this message that do not relate to the official business
of this company shall be understood as neither given nor endorsed by it.
--
//www.freelists.org/webpage/oracle-l
Other related posts:
- » Discrepancy between v$sysmetric and v$sysstat - Allen, Brandon
