les_04-准则-警告-基准( 36 ).ppt

Metrics, Alerts, and Baselines
Objectives
pleting this lesson, you should be able to do the following:
View metrics by using the metrics history views
Create metric thresholds
View alerts
Create metric baselines
Enable adaptive thresholds
Metrics, Alerts, and Baselines
Metric: Rate of change in a cumulative statistic
Alert: Event generated when a monitored metric crosses a threshold
Baseline: Data gathered from a “normal running database” for parison
Limitation of Base Statistics
30 minutes
30 minutes
V$SYSSTATPhysical reads
I/O
I/O
Time
Time
Low rate
High rate
Statistics just keep growing.
2M
2M
Typical Delta Tools
Comparison of statistics at two points in time is needed.
The following tools produce deltas:
Statspack
AWR reports
Customized scripts
Oracle Database 10g Solution: Metrics
Base statistic 1
Redo generation
Metric 1
Metric 2
Redo generation/Tx
Client 1
Client 2
Client 3
Client 4
V$AME
V$SYSMETRIC
V$SESSMETRIC
V$FILEMETRIC
V$EVENTMETRIC
MMON
mit
User rollback
V$WAITCLASSMETRIC
V$SERVICEMETRIC
Benefits of Metrics
DBA
Analyze differences.
Query a metric.
Without metrics
With metrics
Viewing Metric History Information
In memory:
V$SYSMETRIC_HISTORY
V$FILEMETRIC_HISTORY
V$WAITCLASSMETRIC_HISTORY
V$SERVICEMETRIC_HISTORY
V$SESSION_WAIT_HISTORY (last 10 events)
On disk:
DBA_HIST_SYSMETRIC_SUMMARY
DBA_HIST_SYSMETRIC_HISTORY (alerts only)
DBA_HIST_SESSMETRIC_HISTORY
DBA_HIST_SYSTEM_EVENT (cumulative)
DBA_HIST_FILEMETRIC_HISTORY (alerts only)
DBA_HIST_FILESTATXS (cumulative)
DBA_HIST_WAITCLASSMET_HISTORY (alert)
Viewing Detailed Information for a Metric
Statistic Histograms
30 minutes
V$SYSSTATPhysical reads
I/O
Time
High rate
1
2
4
8
16
32
64
128
256
512
…
222
Numberofwaits
Timeperiod
V$FILE_HISTOGRAM
Is it areal problem?
5500