I was looking for some input about AWS auto-scaling. I'm running into an issue where AWS is scaling out twice in a row even though there's only a pretty small traffic increase.
I've got an alarm set up to add 2 tasks when my RequestCountPerTarget exceeds a threshold. These should more than cover the increased traffic. The interval on the alarm is for 3 datapoints (so within 3 minutes).
However it seems to take about 5 minutes for AWS to add the tasks. I suspect this is the problem - can it fire the alarm again, while the 2 tasks are still being added / registered? (before it's reached a "steady state")
And if so, is there any better alternative to increasing the 3-minute interval? (I'm concerned we won't react as quickly to traffic spikes.)
Appreciate any input!