5.7

Option Description
Eliminate non-linear functions for
data sets less than number of data
points points
Specifies that the best fit function use the linear function when the number
of data points is less than this setting.
Force Forecast Method
Forces the forecast function for trend analysis. The linear method provides
a true representation of the range.
Trend and Forecast Data Filter
Trend collected data with no filtering
Prevents the use of outlier detection or smoothing.
Apply one or more smoothing filters
prior to trending
Activates outlier detection and smoothing options.
Smooth data prior to fit
Activates smoothing. You can select the smoothing method.
n
EWMA (Exponentially Weighted Moving Average) calculates each
smooth data point as a weighted mean of the preceding data points such
that more weight or importance is given to data points closer to it.
n
GWMA (Geometrically Weighted Moving Average) applies a
weighted geometric mean instead of a weighted arithmetic mean.
GWMA is more stable amidst fluctuations than EWMA.
n
Lifetime Average calculates each smooth data point as a simple
arithmetic mean of the preceding data points.
Detect outliers and plateaus using
progressive fit analysis
Activates outlier detection.
Outlier detection variance limit
Sets an error threshold against the progressive fits that are part of the
algorithm to detect outliers.
Outlier percentage of data allowed
before giving up
Sets the threshold for the percentage of data points marked as outliers. This
is useful when vCenter Operations Manager detects too many outliers and
the data set becomes too small to continue outlier detection. The calculation
for the number of data points is (outlier percentage limit x total number of data
points) rounded up to the next whole number. Outliers that appear in the
Average Virtual Machine Capacity view depend on the VM Count Capacity
compound metric and outliers might exceed the threshold in this global
setting because each outlier indicates at least one outlier in the dependent
metrics.
4 Click OK or Finish to save your settings, or select another option to configure.
5 Click Done to close the Configuration dialog box.
The Trend and Forecast Analysis settings are reconfigured.
Example: Examples of Trend and Forecast Analysis Global Settings
These examples show how some of the settings are used.
Forecast Functions - Use
up to data point number
data points for trend and
forecast
You might set the viewing default time window to six but use more data points
for this setting to calculate the trend.
Outlier percentage of
data allowed before
giving up (outlier
detection)
If you have eight data points and you set the percentage to 10,
vCenter Operations Manager takes the final value of .8 and rounds it up to 1.
Chapter 8 Set How Data Appears in vCenter Operations Manager
VMware, Inc. 105