We have a requirement to calculate the availability of eChannel applications. The main point is that if we are monitoring the URL from two public locations and two private locations. The service availability counted as 100% even if one of the locations is reachable. How do achieve this requirement?
Note: The excluding periods with maintenance windows is pretty much obvious.
I am not sure, need to do some tests, but we need to check if the builtin metrics for Synthetic Availability takes the Outage configuration from the synthetic monitor you are looking for. Example, if you set the alert to happen only if all locations test fails, the availability metric should still at 100%, or if it will show the "real" availability based on the locations results, for example 4 locations, only one is good, the metric will show 100% or 25%?
After confirm the metric behavior, you can then define a SLO for the Synthetic using the default wizard.
If you set the maintenance windows to stop the synthetics execution, you will be good with it. But if the Synthetics are still running, I believe (really not sure), it will "ignore" the window and continues to calculate the percentage as usual (if failures happen during window, it will decrease the availability %).
Correcting my self here, there is a metric that excludes the Maintenance Window periods, builtin:synthetic.browser.availability.location.totalWoMaintenanceWindow.
Thank you for all your thoughts about the subject requirement.
if you set the alert to happen only if all locations test fails, the availability metric should still at 100%, or if it will show the "real" availability based on the locations results, for example 4 locations, only one is good, the metric will show 100% or 25%?
The above is purely related to the alerting mechanism.
I did a try with the maximum aggregation, even though the availability is reducing when 1 of 4 public locations is still reachable.