To create a histogram (breakdown by threshold values).
- In DMI, create a new section type.
We are now in the section-level DMI editor.
- In this example, let's set Display name to
Histogramand Section type to
- On the Data tab, click
Software service, operation, and site data(the default data view).
- Click Histogram.
Histograms are available for most data views. If histograms are unavailable for the selected data view, Histogram is grayed out.
- Select a metric on which you will set thresholds. These thresholds are used to define the buckets of your histogram.
In this example, select Availability (total) in the Availability section. (Search for "Availability (total)" in the search box.)
- In the Ranges box for Availability (total) (the selected dimension), enter the threshold values for the histogram.
Use semicolons to separate the values. You can specify up to 20 thresholds.
In this example, we use
0;50;60;75;85;95;99to say that we want to break the histogram down at those thresholds.
- Click Metrics and select Total bytes so we can show the total bytes for each bucket in the histogram.
- Click Display report to see it in action.
Each row shows a bucket defined by a pair of thresholds you entered in the Ranges box, and the total bytes for that row's bucket.
For example, when total availability was between 95 percent and 99 percent, the total bytes was 884 MB.
The following rules apply when defining a histogram:
Suffixes are permitted to express units of time (s, m, h, d), “%” for percentage metrics, and suffixes expressing quantities (k, M, G).
If zero is specified as a threshold, the first range will contain only zero values. If a value greater than zero is specified as the first threshold, everything lower than or equal to the first threshold will be in the first range. The subsequent ranges contain values greater than the lower threshold and less or equal to the upper threshold (if any).
In DMI, a histogram applies to input data values, not to result values. A histogram is calculated before data aggregation is done.
Since data is monitored at a defined monitoring interval, only average data is assigned to ranges in histograms. For example, a histogram will not display a single measured value of operation time, but an average operation time, so that all the data from one monitoring interval is assigned to the same range. If the average value does not exceed a particular threshold, it does not mean that no single value exceeded it. This may result in some ranges on the histogram being empty.
If you choose a “unique” type of metric such as Unique users, the histogram will show the number of elements whose value has ever been in a defined range. For example, if a histogram shows the operation time for unique users, with three defined thresholds, each of the four ranges will contain the number of users who have ever had an operation time in each of these ranges. This means that a particular user can be assigned to more than one range.
Additional characters are allowed at the start of the range specifiers:
+ A plus indicates that the threshold is to be treated as an upper boundary.
* An asterisk indicates that the threshold is to be treated as a lower boundary.