Difficulty understanding AWS cloudwatch alarms

Question

I am going through the documentation of CloudWatch alarms https://docs.aws.amazon.com/AmazonCloudWatch/latest/monitoring/AlarmThatSendsEmail.html . There are example scenario tables in the Configuring How CloudWatch Alarms Treat Missing Data section. I am having difficulty understanding what is going on.

In the two last rows why behaviour of Missing and Ignore column is different?

Answer 1

First of all, the last 2 rows are still very different. Although they both have 2 missing data points, the very last row's last data point is a 'X', which is Breaching/bad while the second last rows's last data point is a 'O' which is OK/good. Under the setting of treating missing data as "MISSING"/"IGNORE", the second last row is considered an OK, even if it is missing 2 data points. It is reasonable that MISSING/IGNORE settings are more permissive than BREACHING.

And in the last row, MISSING/IGNORE also have different behaviors. This is because IGNORE is more permissive than MISSING as you can see the IGNORE will "Retain current state". This means your alarm under that circumstance will just stay as is until new data points come in so that it break the current data point pattern.

And the rationale behind the behavior of MISSING in last row is that, although we see a single bad data point, we need more data point to determine the next alarm state to be good/bad, or INSUFFICIENT, if no more data points.