Propagation of execution status with SLA method "Best Child"
The picture below shows the idea behind "Best Child". The best status at a certain point in time is propagated to top and therefore determining the overall SLA status for this point in time. This means in order green is better than red is better than grey the availability is rated according to the best measured status in whole Robot landscape.
8:00 The statuis is grey because we simply do not know it before first script execution.
8:05 Overall status is green because we measured a green execution with robot 1
8:10 Overall status is not red because a few minutes before it was mesured green and we suspect a local problem at Robot2's location
8:30 The green execution from 8:05 is now 2 x schedule.period.seconds in the past, for some reason we detected no second measurement from robot1. As we simply do not know the status from robot1's location any more and robot2 had trouble to execute sucessfully we rate the overall status red.
8:40 Robot2 again confirms the problem => overall still red
8:45 Robot1 delivers a red execution => red + red = red
8:55 Robot2 was able to execute the script => backend is available again => red (robot 1)+ green (robot2) = green
The order "green better than red better than grey = undefined " implicitly includes a situation where 100% green execution do not reflect in 100% SLA availability.
See picture below: At 8:41 we would have expected two more executions from Robot 1 and at 8:56 two more from Robot2. But nothing was measured...why doesn't matter ;o)
Unfortunately we cannot be sure that the system is available if we have no data measured. Therefore we preferred to reduce the availability to the time frames where we have proof of evidence.