Contents IT Operations Management Previous Topic Next Topic Override upper and lower bounds for a metric ... SAVE AS PDF Selected Topic Topic & Subtopics All Topics in Contents Other Share Override upper and lower bounds for a metric You can specify static upper and lower bounds for an Operational Intelligence metric, that override the learned bounds and affect anomaly scores. Before you beginRole required: evt_mgmt_admin About this task Operational Intelligence establishes upper and lower control bounds for metrics using statistical models that learn historical metric data. Upper and lower bounds are then used in the analysis for detecting anomalous CIs. However, a specific metric might have a known typical value for upper or lower bounds, that is based on data in your organization. For example, the value for a CPU metric does not exceed 85%. When such concrete upper or lower bound values are known for a metric, you can configure that metric with static bounds to override the statistical analysis. When a metric value is outside a control bound, the deviation beyond the control bound is normalized by a width measurement that is usually computed from data. Lower and upper width values determine how much a metric value needs to exceed the control bounds to be anomalous. Changing the width that is used to compute the increment above or below control bounds, affects how quickly a high anomaly score is reached. Higher values for lower and upper width, decrease anomaly scores and so decrease the time for a CI to become anomalous. When static values from a metric class are applied, a different method can be used to compute the lower and upper widths, or the widths can be completely overridden. A metric class is where you define custom control bounds and custom widths. Static upper and lower bounds from a metric class override the learned bounds for the metric, and subsequent analysis uses the static bound values. The high level steps to set static values for metric bounds and widths for a specific metric are: Create a metric class in which you define custom values. Associate this metric class with the metric that it applies to. A metric class can be set with a pre-defined class type that determine how static lower and upper bound values are computed and used with the statistical model: Static Centered Intended to be used with metric data that is typically spread symmetrically between the upper and lower bounds. With this setting, the formula that is used to set bounds and width values, ignores the statistical data. In this setting, the lower and the upper widths have the same value. Static Skewed Intended to be used with metric data that is not evenly spread between the upper and lower bounds, and that typically tends to concentrate closer to one of the bounds. The upper and lower bounds in this case, are not symmetric around the data. With this setting, the median of the data is used to separately compute an upper width and a lower width. Other Applies a statistical model other than one of the static models (Static Centered and Static Skewed). With this setting, the distance between the bounds decreases as the values for the upper bound and lower bound are set as follows: Lower bound is the higher value between the learned lower bound and the Lower bound specified in the metric class. Upper bound is the lower of the learned upper bound and the Upper bound specified in the metric class. You can also override the learned lower and upper width values for a metric. These values determine how much a metric value needs to exceed the bounds to be anomalous. Changing the range of values (width) used to compute the increment above or below bounds, affects the speed of reaching a high anomaly score. Procedure Navigate to Operational Intelligence > Metric Classes. In the Metric Class list view, click New. Fill out the Metric Class form, and then click Submit. Field Description Name A unique name for the metric class. Lower Bound Static value that overrides the learned lower bound for the metric. Upper Bound Static value that overrides the learned upper bound for the metric. Upper Bound must be greater than Lower Bound. Class Type Determines how the static bounds are used with the statistical model. Static Centered Static Skewed Other Lower Width Override Applies when Class Type is set to Static Skewed. A static value that overrides the learned lower width. Upper Width Override Applies when Class Type is set to Static Skewed as a static value that overrides the learned upper width. Associate the metric class with a metric type: Navigate to Operational Intelligence > Metric Types. Double-click in the Metric Class column of the metric type that you want to associate the metric class with. Select a metric class and then click Save. On this page Send Feedback Previous Topic Next Topic

Override upper and lower bounds for a metric You can specify static upper and lower bounds for an Operational Intelligence metric, that override the learned bounds and affect anomaly scores. Before you beginRole required: evt_mgmt_admin About this task Operational Intelligence establishes upper and lower control bounds for metrics using statistical models that learn historical metric data. Upper and lower bounds are then used in the analysis for detecting anomalous CIs. However, a specific metric might have a known typical value for upper or lower bounds, that is based on data in your organization. For example, the value for a CPU metric does not exceed 85%. When such concrete upper or lower bound values are known for a metric, you can configure that metric with static bounds to override the statistical analysis. When a metric value is outside a control bound, the deviation beyond the control bound is normalized by a width measurement that is usually computed from data. Lower and upper width values determine how much a metric value needs to exceed the control bounds to be anomalous. Changing the width that is used to compute the increment above or below control bounds, affects how quickly a high anomaly score is reached. Higher values for lower and upper width, decrease anomaly scores and so decrease the time for a CI to become anomalous. When static values from a metric class are applied, a different method can be used to compute the lower and upper widths, or the widths can be completely overridden. A metric class is where you define custom control bounds and custom widths. Static upper and lower bounds from a metric class override the learned bounds for the metric, and subsequent analysis uses the static bound values. The high level steps to set static values for metric bounds and widths for a specific metric are: Create a metric class in which you define custom values. Associate this metric class with the metric that it applies to. A metric class can be set with a pre-defined class type that determine how static lower and upper bound values are computed and used with the statistical model: Static Centered Intended to be used with metric data that is typically spread symmetrically between the upper and lower bounds. With this setting, the formula that is used to set bounds and width values, ignores the statistical data. In this setting, the lower and the upper widths have the same value. Static Skewed Intended to be used with metric data that is not evenly spread between the upper and lower bounds, and that typically tends to concentrate closer to one of the bounds. The upper and lower bounds in this case, are not symmetric around the data. With this setting, the median of the data is used to separately compute an upper width and a lower width. Other Applies a statistical model other than one of the static models (Static Centered and Static Skewed). With this setting, the distance between the bounds decreases as the values for the upper bound and lower bound are set as follows: Lower bound is the higher value between the learned lower bound and the Lower bound specified in the metric class. Upper bound is the lower of the learned upper bound and the Upper bound specified in the metric class. You can also override the learned lower and upper width values for a metric. These values determine how much a metric value needs to exceed the bounds to be anomalous. Changing the range of values (width) used to compute the increment above or below bounds, affects the speed of reaching a high anomaly score. Procedure Navigate to Operational Intelligence > Metric Classes. In the Metric Class list view, click New. Fill out the Metric Class form, and then click Submit. Field Description Name A unique name for the metric class. Lower Bound Static value that overrides the learned lower bound for the metric. Upper Bound Static value that overrides the learned upper bound for the metric. Upper Bound must be greater than Lower Bound. Class Type Determines how the static bounds are used with the statistical model. Static Centered Static Skewed Other Lower Width Override Applies when Class Type is set to Static Skewed. A static value that overrides the learned lower width. Upper Width Override Applies when Class Type is set to Static Skewed as a static value that overrides the learned upper width. Associate the metric class with a metric type: Navigate to Operational Intelligence > Metric Types. Double-click in the Metric Class column of the metric type that you want to associate the metric class with. Select a metric class and then click Save.

Override upper and lower bounds for a metric You can specify static upper and lower bounds for an Operational Intelligence metric, that override the learned bounds and affect anomaly scores. Before you beginRole required: evt_mgmt_admin About this task Operational Intelligence establishes upper and lower control bounds for metrics using statistical models that learn historical metric data. Upper and lower bounds are then used in the analysis for detecting anomalous CIs. However, a specific metric might have a known typical value for upper or lower bounds, that is based on data in your organization. For example, the value for a CPU metric does not exceed 85%. When such concrete upper or lower bound values are known for a metric, you can configure that metric with static bounds to override the statistical analysis. When a metric value is outside a control bound, the deviation beyond the control bound is normalized by a width measurement that is usually computed from data. Lower and upper width values determine how much a metric value needs to exceed the control bounds to be anomalous. Changing the width that is used to compute the increment above or below control bounds, affects how quickly a high anomaly score is reached. Higher values for lower and upper width, decrease anomaly scores and so decrease the time for a CI to become anomalous. When static values from a metric class are applied, a different method can be used to compute the lower and upper widths, or the widths can be completely overridden. A metric class is where you define custom control bounds and custom widths. Static upper and lower bounds from a metric class override the learned bounds for the metric, and subsequent analysis uses the static bound values. The high level steps to set static values for metric bounds and widths for a specific metric are: Create a metric class in which you define custom values. Associate this metric class with the metric that it applies to. A metric class can be set with a pre-defined class type that determine how static lower and upper bound values are computed and used with the statistical model: Static Centered Intended to be used with metric data that is typically spread symmetrically between the upper and lower bounds. With this setting, the formula that is used to set bounds and width values, ignores the statistical data. In this setting, the lower and the upper widths have the same value. Static Skewed Intended to be used with metric data that is not evenly spread between the upper and lower bounds, and that typically tends to concentrate closer to one of the bounds. The upper and lower bounds in this case, are not symmetric around the data. With this setting, the median of the data is used to separately compute an upper width and a lower width. Other Applies a statistical model other than one of the static models (Static Centered and Static Skewed). With this setting, the distance between the bounds decreases as the values for the upper bound and lower bound are set as follows: Lower bound is the higher value between the learned lower bound and the Lower bound specified in the metric class. Upper bound is the lower of the learned upper bound and the Upper bound specified in the metric class. You can also override the learned lower and upper width values for a metric. These values determine how much a metric value needs to exceed the bounds to be anomalous. Changing the range of values (width) used to compute the increment above or below bounds, affects the speed of reaching a high anomaly score. Procedure Navigate to Operational Intelligence > Metric Classes. In the Metric Class list view, click New. Fill out the Metric Class form, and then click Submit. Field Description Name A unique name for the metric class. Lower Bound Static value that overrides the learned lower bound for the metric. Upper Bound Static value that overrides the learned upper bound for the metric. Upper Bound must be greater than Lower Bound. Class Type Determines how the static bounds are used with the statistical model. Static Centered Static Skewed Other Lower Width Override Applies when Class Type is set to Static Skewed. A static value that overrides the learned lower width. Upper Width Override Applies when Class Type is set to Static Skewed as a static value that overrides the learned upper width. Associate the metric class with a metric type: Navigate to Operational Intelligence > Metric Types. Double-click in the Metric Class column of the metric type that you want to associate the metric class with. Select a metric class and then click Save.