KeyConfigure’s internal reports are extremely powerful and flexible. Time Sets are a feature you may not have noticed though, even if you are fully versed in using Time Ranges, parameters, filters, adhoc targets, multiple views, and custom columns. Essentially, Time Sets let you run a report on a long Time Range such as a month, but split apart usage occurring in different parts of each week or day.
First, let’s look at a simple Histogram Logins (DIV) report. Here, we’ve run it on a Time Range of three weeks, and we’re looking at the SAN-Proto division.
The histogram is already fairly informative – we can see the maximum concurrent number of logins over this Time Range, and we can begin to understand patterns of logins. The Time Range is short enough that we can see a peak every day, and we can tell that weekday peaks are higher than weekend peaks. But let’s imagine that there are very specific times during the week that we want to separate out from the rest.
For example, what sort of logins are we seeing in the early morning? We could choose a single day and run the report on a few hours from that day. But we want to understand more globally what logins are like during those hours but over the course of the entire 3 weeks. To do this we will configure Time Sets.
Having brought the Time Sets window to the foreground, a right-click allows us to select “New Time Set” and then another right-click in the Time Periods list brings up the “Time Period” configuration window. Each time Time Period specifies the days/hours it includes. Note that a single Time Set can be comprised of multiple disjoint Time Periods (e.g. Monday 8-9AM AND Tuesday 2-3PM).
Our list of Time Sets above shows how we have split out Weekends, normal working hours, and various before/after hour Time Sets. The Blue dashes in the Coverage column give a visual preview of what parts of the week are included. Notice the misconfiguration – the Time Set labeled “Weekdays Late” actually includes 5:30PM-9:30PM on every day, not just weekdays.
Once we have defined some Time Sets, we can apply them to a new report by selecting them in the Time Periods pane of the Report Builder window:
Our original histogram report is now essentially broken out into several sub-reports. In the example below, for each lab division, the detail rows on the right show the peak for each Time Set and when the line is selected, we see the histogram for this period:
Concurrent logins are highest during normal daytime hours. They are lower in early/late hours, and even lower on weekends. We could have guessed this, and we could have more or less seen this trend by looking at the original graph – but imagine following these same steps using much more specific Time Sets. This can be a very powerful way to analyze usage data when you are looking for specific data.
This is in fact exactly what the K2 Administrator at Evergreen State College did. A Budget reduction forced Evergreen to cut back on lab hours but students were protesting. Using Time Sets, the K2 Administrator looked at what the effect would be of for example closing a certain lab 2 hours earlier. They could see the hours during the week when the labs were almost empty, and demonstrate to students that a schedule change would have a low impact. You can read more about how they reduced operational costs in our IT Resource Management Case Study.