5.7

Outlier Detection and Smoothing Filters in vCenter Operations Manager
vCenter Operations Manager provides data filtering techniques to address outliers and temporary fluctuations
in data points. Outlier detection and smoothing remove anomalies or sudden changes in patterns during
maintenance or other cycles that might mislead the trend and forecast calculations.
When you apply smoothing, vCenter Operations Manager applies a smoothing function to data points and
calculates the trend using the smooth data set. The value of a data point in a graph is affected by the value of
the preceding data points. vCenter Operations Manager assigns a weight or priority to each data point and
that weight determines the contribution of the data point to a smoothed value.
Activating different combinations of outlier detection and smoothing generates different results.
Table 8-2. Outlier Detection and Smoothing Results
Data Filter Result
No data filter Shows the original data points and trend.
Outlier detection Shows the original data points, trends, and outliers. In the
views, outliers appear as hollow circles.
Smoothing Shows the smooth data points and trends.
vCenter Operations Manager does not show the original data
points.
In a table view, vCenter Operations Manager identifies the
smooth data points with (S).
Outlier detection and smoothing Shows the smooth data points, trends, and outlier data.
vCenter Operations Manager does not show the original data
points. In this case, outlier detection occurs first followed by
smoothing.
If the data involves 10 data points, outlier detection
eliminates two data points as outliers and smoothing occurs
on the remaining eight data points.
You can set the trend, outlier detection, and smoothing filters in the global settings.
NOTE The Average Virtual Machine Capacity view relies on the VM Count Capacity compound metric for
data. Because the outliers in this view indicate outliers in at least one of the dependent metrics for the time
stamp, the view might exceed the threshold for the percentage of data points marked as outliers in the Outlier
percentage of data allowed before giving up policy setting. To investigate the resource dimension that might
cause the outlier, look at the Cluster or Host Capacity Usage view.
The settings for outlier detection and smoothing are located in the 6 Configure forecast and trends section of
each policy.
Modify an Existing Policy
You can modify a previously configured policy to improve the way vCenter Operations Manager interprets
and presents data in your environment.
An administrator can modify any policy in vCenter Operations Manager at any time. The changes that the
administrator applies affect all users. Therefore, the administrator must notify the users who are working with
vCenter Operations Manager about any new settings that are applied to the policies.
Associating Policies
with Groups
You can associate policies to specific groups.
NOTE You can associate only one policy per group, while one policy can be
associated with many groups.
VMware vCenter Operations Manager Getting Started Guide
106 VMware, Inc.