Average Response Time Calculation in Reports

Response time is an important monitoring metric available in your monitors and reports: it shows the response time of your server measured by monitoring checks.

In most of your in-dashboard reports, you can choose to show in the report data of your monitors per individual location, or for all locations.

Click the link in the topmost right corner of your report screen to select the scope of your report.

SLA Uptime Report

You can then select one of the below options:

  • All Locations (to show data of your monitors across all locations)
  • All with Locations (to show data of your monitors grouped by location, as well as data across locations)
  • Any individual location in the list (to show data for your selected location)


Average response time in reports is calculated differently for an individual location and across all locations.

Average response time calculation methods used in reports are detailed further in this document.

Average Response Time Calculation for Individual Location

Note: the examples for different average response time calculation methods described herein and further in this document use one-hour data reporting (aggregation) interval.

If you select any location from the list, your report will show average response time of your monitors for the selected location.

SLA Uptime Report Location

For all types of external monitoring (Uptime, Transaction and Full Page Load monitors), an individual location’s average response time over a reporting (aggregation) interval is calculated by the following method:

  • The average response time of monitoring checks that have returned OK result out of all monitoring checks of the location completed within the reporting (aggregation) interval:[Average response time individual location] = [Average response time of location’s checks that have returned OK result] = [Sum of response time of location’s checks that have returned OK result] / [Number of all location’s checks that have returned OK result]

The number of checks accomplished within the reporting (aggregation) interval depends on the monitoring check frequency that you have set for a particular location in your monitor’s settings.

Settings_

The number of checks within the reporting period is equal to:

[Number of checks within reporting interval] = [Reporting period]) / [Check frequency interval]

So if e.g. you have configured check frequency for your monitor from US-MID location to be 3 min, the total number of checks within the one-hour interval will be equal to:

60/3 = 20

Example: average response time calculation for an individual location for Uptime monitors

The below example illustrates the described individual location average response time calculation method for an Uptime monitor.

The example uses one-hour reporting interval.

Report-screenshot-6

In the example, the user has configured monitoring from 4 locations, with the check frequencies as follows:

  • LOC 1: 1 min
  • LOC 2: 2 min
  • LOC 3: 3 min
  • LOC 4: 5 min
Note that the checks from locations that have returned OK result show the response time values (msec).

Calculating by the above described method, for the one-hour reporting (aggregation) interval:

  • Location 1: [Sum of response times of checks that have returned OK result] / [Number of checks that have returned OK result] = 280/40 = 7 m/sec
  • Location 2: [Sum of response times of checks that have returned OK result] / [Number of checks that have returned OK result] = 109/21 = 5.19 m/sec
  • Location 3: [Sum of response times of checks that have returned OK result] / [Number of checks that have returned OK result] = 64/13 = 4.92 m/sec
  • Location 4: [Sum of response times of checks that have returned OK result] / [Number of checks that have returned OK result]
  • = 76/10= 7.6 m/sec

Example: average response time calculation for individual location for Transaction/FPL monitors

The below example illustrates the described individual location average response time calculation method for a Transaction (or, similarly, a Full Page Load) monitor.

The average response time calculation method is the same as for Uptime monitors.

The example uses one-hour reporting interval.

Report-screenshot-7

 

In the example, the user has configured monitoring from 4 locations, with the check frequencies as follows:

  • LOC 1: 5 min
  • LOC 2: 10 min
  • LOC 3: 15 min
  • LOC 4: 20 min
Note that the checks from locations that have returned OK result show the response time values (msec).

Calculating by the above described method, for the one-hour reporting (aggregation) interval:

    • Location 1: [Sum of response times of checks that have returned OK result] / [Number of checks that have returned OK result] = 63/9 = 7 m/sec
    • Location 2: [Sum of response times of checks that have returned OK result] / [Number of checks that have returned OK result] = 28/6 = 4.67 m/sec

  • Location 3: [Sum of response times of checks that have returned OK result] / [Number of checks that have returned OK result] = 9/2 = 4.5 m/sec
  • Location 4: [Sum of response times of checks that have returned OK result] / [Number of checks that have returned OK result] = 8/1 = 8 m/sec

All with Locations

If you select “All with Locations” from the list, your report will show average response time of your monitors grouped by location for your specified time range.

 

SLA Uptime Report All With Locations

“All with Locations” average response time uses the same calculation method for individual locations as described above: it just consolidates your monitors by location in the report.

Notice that “All with Locations” option shows also the “All Locations” data, which is available also in a separate “All Locations” option in the report scope selection combo box, as described further in this document.

Average Response Time Calculation for All Locations

If you select “All Locations” from the list, your report will show average response time of your monitors across all locations for the set time range.

For all types of external monitors (Uptime, Transaction and Full Page Load) the “All Locations” average response time is calculated by the following method:

– As different locations for a monitor may have different check frequency intervals set by the user (see Uptime Monitoring, Transaction Monitoring and Full Page Monitoring), the smallest check frequency interval among all locations is taken to determine the response time across all locations for each such interval. If, for example, the user has set monitoring check frequencies to be 1, 2, 3 and 5 min, then 1 min is the smallest check frequency interval.

– For each of the smallest check frequency intervals, locations that do not run check during that interval are excluded when determining the “All Locations” average response time.

– For each of the smallest check frequency intervals, if there is at least one OK result returned by a location, then the “All Locations” uptime check result for that interval is assigned OK status. If all locations running check within that interval return NOK, the “All Locations” uptime check result for that interval is assigned NOK status (see Uptime Calculation in Reports).

– For every reporting interval (1 hour in our example), the “All Locations” average response time result is then calculated as follows:

– The smallest frequency interval checks that have returned NOK as the “All Locations” uptime status are excluded from average response time calculation for the reporting interval.

– For each of the smallest check frequency intervals that have returned OK as the “All Locations” uptime status, the minimum response time reported by a location during that interval is determined as “All Locations” response time for that interval.

– The “All Locations” average response time for the reporting interval is then calculated as an average of the minimum response times of the smallest check frequency intervals that have returned OK as an uptime status.

[Average response time “All Locations”] = [Average of minimum response times of the smallest frequency interval checks that have returned OK uptime status]
The number of checks within the reporting period is equal to:

[Number of Checks] = [Reporting interval] / [Smallest check frequency interval]

That means that if, for example, the smallest check frequency interval for a monitor among all locations is 1 min, then the number of smallest check frequency intervals in the reporting interval of one-hour would be:

60/1 = 60

If the smallest check interval is 2 min, the number of smallest check frequency intervals in the reporting interval would be:

60/2 = 30

Example: calculation of “All Locations” average response time for Uptime monitors

The below example illustrates the described “All Locations” average response time calculation method for Uptime monitors. The example uses one-hour reporting interval.

Report-screenshot-8

In the shown example, the user has configured monitoring for an uptime monitor from 4 locations, with the check frequencies set for locations as follows:

  • LOC 1: 1 min
  • LOC 2: 2 min
  • LOC 3: 3 min
  • LOC 4: 5 min

The smallest check frequency interval is 1 min, so “All Locations” uptime status (OK or NOK) is determined every 1 min during the one-hour reporting interval, as shown in the example.

The number of one-minute check intervals within the reporting interval (one hour in our example) will be:

60/1 = 60

As shown in the example, e.g. for the 6th minute the check has returned:

  • LOC 1: OK
  • LOC 2: NOK
  • LOC 3: OK
  • LOC 4: no check carried out, as 5-min check frequency interval is set for this location
Note that the checks from locations that have returned OK result show the response time values (msec).

As there is at least one OK result returned by a location, “All Locations” uptime status is assigned OK for this smallest check frequency interval, as shown in the example.

The response time per location for the 6th minute check in the example is:

– LOC 1: 5 msec

– LOC 2: N/A, as the check result is NOK

– LOC 3: 12 msec

– LOC 4: N/A, as no check run

As it can be seen, the minimum response time returned by a location (locations that have not run check or returned NOK omitted) is 5 msec returned by Location 1.

Now, it we look e.g. at the results of the 10th minute check, we can see that:

  • LOC 1: NOK
  • LOC 2: NOK
  • LOC 3: no check carried out, as 3-min check frequency interval is set for this location
  • LOC 4: NOK

As there is not at least one OK result returned by a location within this check interval, “All Locations” uptime status is assigned NOK for this smallest check frequency interval. Therefore, this 1-min interval will be omitted in calculation of “All Locations” average response time for the reporting interval (1 hour in the example).

Now, if we calculate the average of minimum response times of the smallest check intervals (1 min in our example) that have returned OK as “All Locations” uptime result within the reporting interval (1 hour in our example), for the shown example it will be:

[Average response time “All Locations”] = 261 / 36 = 7.25 msec

Note: if due to a technical reason a location delays reporting the check data, the location will be excluded from calculations in that interval.

Example: calculation of “All Locations” average response for Transaction/FPL monitors

The below example illustrates the described individual location average response time calculation method for a Transaction (or, similarly, an FPL) monitor.

The average response time calculation method is the same as for uptime monitors.

The example uses one-hour reporting interval.
Report-screenshot-9

In the example, the user has configured monitoring for a Transaction monitor from 4 locations, with the check frequencies set for locations as follows:

  • LOC 1: 5 min (currently, the minimum available frequency for Transaction and Full Page Load monitors)
  • LOC 2: 10 min
  • LOC 3: 15 min
  • LOC 4: 20 min

The smallest check frequency interval is 5 min, so “All Locations” uptime status (OK or NOK) is determined every 5 min during the one-hour reporting interval, as shown in the example.

The number of five-minute check intervals within the reporting interval (one hour in our example) will be:

60/5 = 12

As shown in the example, e.g. for the 20th minute the check has returned:

  • LOC 1: OK
  • LOC 2: OK
  • LOC 3: no check carried out, as 15-min check frequency interval is set for this location
  • LOC 4: NOK
Note that the checks from locations that have returned OK result show the response time values (msec).

As there is at least one OK result returned by a location, “All Locations” uptime status is assigned OK for this smallest check frequency interval, as shown in the example.

The response time per location for the 20th minute check in the example is:

– LOC 1: 14 msec

– LOC 2: 16 msec

– LOC 3: N/A, as no check run

– LOC 4: N/A, as the check result is NOK

As it can be seen, the minimum response time returned by a location (locations that have not run check or returned NOK omitted) is 14 msec returned by Location 1.

Now, it we look e.g. at the results of the 35th minute check, we can see that:

  • LOC 1: NOK
  • LOC 2: no check carried out, as 10-min check frequency interval is set for this location
  • LOC 3: no check carried out, as 15-min check frequency interval is set for this location
  • LOC 4: no check carried out, as 20-min check frequency interval is set for this location

As there is not at least one OK result returned by a location within this check interval, “All Locations” uptime status is assigned NOK for this smallest check frequency interval. Therefore, this 5-min interval will be omitted in calculation of “All Locations” average response time for the reporting interval (1 hour in the example).

Now, if we calculate the average of minimum response times of the smallest check intervals (5 min in our example) that have returned OK as “All Locations” uptime result within the reporting interval (1 hour in our example), for the shown example it will be:

[Average response time “All Locations”] = 59 / 7 = 8.43 msec

Note: if due to a technical reason a location delays reporting the check data, the location will be excluded from calculations in that interval.

“All Locations” Average Response Time Calculation for Different Reporting Intervals

Different reports may be using different reporting (aggregation) intervals.

For any reporting interval the “All Locations” average response time calculation is done by the same method as described above. The count of monitoring checks and OK results is taken for the reporting interval used in the report (e.g. daily), in the same way as in the above described example of one-hour reporting interval.