Oracle11gEnhancementsinAWRBaselines - 数据库编程

tics to define the alert thresholds specific to the baseline.

Baselines capture metric values, which the database will then compare against current performance metrics to judge how current performance measures up against performance during a known good period. If there’s a serious discrepancy—that is, if the expected values are very different from the actual present statistics—the database will issue a performance alert. Whether you use a manually computed static baseline or a baseline automatically computed over the system moving window, the baseline values are compared to present statistics to see if an alert is justified. Adaptive thresholds are so named because the thresholds aren’t fixed, but vary according to the conditions in the database—they adapt to the type and amount of the workload.

The database computes statistics from the system moving window according to the BSLN_MAINTAIN_STATS_SCHED schedule.

The metric statistics that you capture over a baseline enable you to set thresholds for comparing baseline statistics to current activity. You can use three different threshold types, as explained here:

Significance level Thresholds based on significance level use statistical significance todetermine whether the current levels observed are unusual compared to baseline values, thus meriting an alert. For example, if you set the significance level to 0.99 for a metric, the alert threshold will be setwhere 1 percent of the observed metric values are outside the value set for the metric. The database will thus issue an alert when 1 percent of the metric values are different from the expected metric value. Note that the higher the significance level, the fewer the number of alerts that will be issued by the database. For example, a significance level of 0.9999 would cause fewer alerts to be raised than a significance level of 0.99.
Percentage of maximum An alert is generated if the observed metric is at or above a percentage of the maximum that you specified. For example, if you specify 120 as the percentage of maximum where the maximum value captured by the baseline is 1000, the database will issue an alert when the observed metric crosses 1200, which is 120 percent of the maximum (1000).
Fixed values Fixed values are standard Enterprise Manager fixed thresholds,which the database compares with the actual metrics. The DBA sets the fixed values, without the need for any AWR baselines.

Oracle Database 11g fully integrates the selection of adaptive thresholds for performance metrics with the AWR baselines, with the baselines serving as the source of the metrics. The database determines the alert thresholds by examining the metric statistics captured over the baseline time period. Thus, the database sets the thresholds based on data provided by the database itself, and you don’t have to know any system-specific metrics. The database sets the thresholds based on system data itself and some metadata provided by you. Using the Enterprise Manager, you can choose a starter set of thresholds based on either the OLTP or the Data Warehouse workload profile.

Once you select the appropriate workload profile, the database will automatically configure and maintain the adaptive thresholds based on the default SYSTEM_MOVING_WINDOW baseline. The adaptive thresholds will cover all metrics suitable for the chosen workload profile. Once you configure the adaptive thresholds, you can edit the thresholds levels. When you’re editing the threshold levels, Oracle recommends that you set the initial signi

Oracle11gEnhancementsinAWRBaselines(七)