Explanation of capacity provisioning and capacity projection

Document ID : KB000023054
Last Modified Date : 14/02/2018
Show Technical Document Details

Questions:

How is the average provisioning/projection information derived or calculated?

What baseline is used for the Health Report Capacity Planning section?

Which elements should appear in the Capacity Projection section of a Health report?

 

Answers:

The Health report Capacity Planning section is based on the baseline + report period.

Capacity planning requires an analysis of trends in your network. By default, eHealth shows you a utilization trend based on a baseline period of six weeks. To develop a valid trend line, it must poll your network elements for the length of the baseline period.

Any element that has an increasing Days To Threshold in the Situations to Watch section should be included in the Capacity Projection. This is the default. To override the default, a Report Developers License is required. Please see:

How to force the Health Capacity planning report to include elements with a negative slope 

Health maintains analyzed data about your resources for the baseline period. The baseline period is one of three rolling periods that projects backward in time from the day the report is run.

  • For a daily Health report, the baseline is 6 weeks (42 days) by default.
  • For a weekly Health report, the baseline is 13 weeks by default.
  • For a monthly Health report, the baseline is 12 months by default.

eHealth reports compare the following:

  • Hourly information for the report period and baseline data for the same hour of the day
  • Daily information for the report period and baseline data for the same days of the week
  • Monthly information for the report period and baseline data for the same months

The $NH_HOME/reports/health/healthStyles.sds file shows the Capacity Projections being calculated with the following data. As you can see, the predicted value is added to the slope value and multiplied by the number of days necessary.

cell 4 {
dataType float
formatStyle stdFloat3
value (variableValue(predictedValue) + (variableValue(slope) * (cond ($(_isMonthly),90.0, cond($(_isWeekly),30.0,30.0)))))
}

cell 5 {
dataType float
formatStyle stdFloat3
value (variableValue(predictedValue) + (variableValue(slope) * (cond ($(_isMonthly),180.0, cond($(_isWeekly),90.0,60.0)))))
}

cell 6 {
dataType float
formatStyle stdFloat3
value (variableValue(predictedValue) + (variableValue(slope) * (cond ($(_isMonthly),360.0, cond($(_isWeekly),270.0,90.0)))))
}

sortOrder 1
sortDescending Yes

The slope is based on least squares regression.

  • The xAxis datapoint is the UTC timestamp representing the time.
  • The yAxis datapoint is the daily value for the variable.
  • For a daily Health report: 30, 60, and 90 days
  • For a weekly Health report: 30 days, 90 days, and 9 months
  • For a monthly Health report: 90 days, 6 months, and 1 year

 

Additional Information:

Explanation of least squares regression
Visual representation of least squares regression