Dealing with gaps in the Power Events Data

Ideally when an eiPower report is run there is a continuous set of power events for all computers covering the entire period between the report Start and End dates. However this is not always the case and eiPower Saver Solution need to be able to handle gaps in the power event data from client computers.
Typically gaps in the power event data are cause for the following reasons;

  1. The Altiris Agent service was not running.
  2. The power events from the agent were not processed by the server due to a licensing exceeded exception.
  3. Old Power Event were purged due to a scheduled purge policy.
  4. The eiPower Agent was not installed or the computer did not exist for period of the reporting period.

While small gaps in the power events data are not uncommon and are not a problem, large gaps can affect the accuracy of the calculations. Reasons 1-2 typically lead to small gaps in the data that will generally not have a significant impact on the accuracy of the overall calculations. Reason 3-4 generally results in large gaps in the data that can completely change the report results. This is felt most when running the graphs that use a comparison to the baseline.

A number of the eiPower Saver reports compare the energy usage, cost and CO2 emission figures for one period (typically called a baseline period) against another period (typically called the comparison period). The baseline period is typically set to be the period immediately after the eiPower Agent was installed onto client computers but before any power policies were applied. The idea is that the power events data collected during the baseline period represent the energy usage patterns for each computer before power policies were applied. Later this data can be compared to a period after power policies have been applied in order to determine how much power has been saved by implementing eiPower.

In order to compare the energy usage, cost and CO2 emissions values for one period against another period it is essential that full sets of data for the same number of computers are compared and that both periods be the same length of time. Additionally it is recommended that the start date for each period be the same day of the week. This is because the energy usage follows a weekly pattern, high during weekdays and lower on weekends, and it is more accurate if like weekdays are compared.

A common problem occurs when new computers enter an organisation or existing computers get replaced at some time after the end of the baseline period. When this occurs there is no baseline data for this new computer and we in effect have a gap in the power events data covering the entire baseline period. This gaps will have a large affect on the calculated energy usage, cost and CO2 emissions values for the baseline period.
Gaps in the data can be dealt with in the following ways;

  • Use average values instead of total values for energy usage, cost and CO2 emissions. Doing this means that gaps in the data are not really a problem because an average is only calculated on the data we actually do have. Many of the reports in eiPower will show the both a total and average figures. Of course it is still necessary to have some data for enough computer in order to calculate an average this is representative of the computers in the report.
  • Assume that the computer was always powered ON during the gap periods. For the big gaps in the baseline data (as described above) assuming the computer was on is a fairly accurate assumption. This is because during the baseline period power policies have not yet been applied and in many organisations this means that computers are left on for the majority of the time.
    Using this assumption it is possible to run a comparison report even if a baseline was never established. Lets says that you wanted to get an idea of savings achieved for last month (1st – 30th April 2010) but that a baseline was never established. In this case you could run the Comparison report with the baseline period can be set to a period when you know that there is no power event data at all (e.g. 1st – 30st January 1900) and the comparison period set to 1st April 2010. When the report is run with this baseline date it will not have any data for the baseline period and hence will assume that all computers were always on during the this period. Although not 100% accurate it will give a rough idea of the amount of power saved.
  • Assume that the computer was always off during the gap period. Generally this is not as accurate as assuming that the computers is on.

In the eiPower reports the parameter called KWH Consumption allows you to specify where the report assumes the computer was powered On or Off during the gaps. Report Measured Usage means that the computer is assumed to be Off, whereas the Report Maximal Usage assumes the computer was On.