DOP-C02 Practice Question 47: Alarm Missing Data Treatment
AWS Certified DevOps Engineer – Professional (DOP-C02) · Test 5 · question 7 of 10
A nightly batch job publishes a custom ProcessingErrors metric to Amazon CloudWatch, but only emits a data point while it is actually processing records. An alarm on that metric is meant to page the on-call engineer when the error count is greater than zero. Instead, the alarm pages every night during the long stretch when the job is idle and no data points are published at all. The alarm's missing-data treatment is currently set to breaching.
The team wants the alarm to page only on a real breaching data point, to show OK on the dashboard while the job is idle, and never to hold a stale state from earlier in the day.
What should the team change?
Show answer and explanation
Answer:
CloudWatch documents the four treatments and gives this exact scenario as the example for choosing notBreaching: "for a metric that generates data points only when an error occurs, such as ThrottledRequests in Amazon DynamoDB, you would want to treat missing data as notBreaching." With notBreaching, "missing data points are treated as 'good' and within the threshold", so the idle hours evaluate to OK, which is both the no-page requirement and the green dashboard the team asked for. A real data point above zero still breaches and still pages, because real data points in the evaluation range are always used before missing ones are filled in.
Option A silences the pages but keeps stale state. With ignore, "the current alarm state is maintained", so if the job's last real data point was breaching, the alarm sits in ALARM all night, which the stem explicitly rules out.
Option C keeps the setting that is causing the problem. With breaching, CloudWatch fills the missing points as breaching; five out of five merely delays the false page by a few periods during a stretch where every point is missing.
Option D stops the paging but shows the wrong thing. With missing, "if all data points in the alarm evaluation range are missing, the alarm transitions to INSUFFICIENT_DATA", grey on the dashboard rather than the OK the team requires. It is also the default, so the alarm's behaviour would just be the out-of-the-box one.
Reference: Configuring how CloudWatch alarms treat missing data