M2.2 module quiz
Ten questions. 80% to pass (8 of 10). Open book, unlimited retakes.
Answers are collapsed under each question. Answer first, then check.
Q1
The Recent Activity tab differs from the Metrics drawer because it reads:
A. Refreshes its data far more frequently
B. Cloud control-plane events rather than utilization metrics
C. Covers only Kubernetes workloads
D. Reads billing export data rather than live telemetry data
Show answer
Correct: B. CloudTrail, Cloud Logging and Azure Activity Log. It answers “has anyone touched this” where metrics answer “how hard is it working”, and the two together distinguish an abandoned resource from a quiet but owned one.
Q2
Activity sync runs:
A. Daily at 20:50 UTC
B. Every 6 hours, on the discovery cycle
C. Hourly, on the hour
D. In real time, as events arrive
Show answer
Correct: A. Fetching the previous day’s logs. Higher frequency would hit cloud log-API rate limits, and cost optimization tolerates roughly 24 hours of staleness.
Q3
A recommendation shows no savings figure and the rule appears in the pricing-gap DLQ. This means:
A. The recommendation is invalid and should therefore simply be dismissed outright
B. The resource was deleted between the scan and the render
C. The rule needed pricing data that was unavailable, so it surfaced as a coverage gap
D. The customer lacks a billing connection for that account
Show answer
Correct: C. Concrete-or-abstain: a rule that cannot price its finding emits nothing rather than a guess. The DLQ entry makes the gap observable, and it can be replayed once the rate lands.
Q4
The periodic-workload trap in the Metrics drawer is:
A. A workload that spikes predictably reads as low-average and looks idle
B. Metrics are sampled too frequently to be meaningful
C. Memory is simply not collected without the CloudWatch agent installed
D. Percentiles are unavailable for that metric family
Show answer
Correct: A. Average-based rules miss the shape. This is why reading the drawer before acting on a rightsizing finding matters: a batch job at 4% average CPU may be at 95% for the hour that matters.
Q5
Realized savings usually fall short of projected savings because:
A. The projection is computed on billed cost rather than on the rack rate
B. Mid-month action, cascading effects and partial adoption all erode the figure
C. Providers change their prices mid-month
D. The rule systematically over-counts hours
Show answer
Correct: B. Acting on the 20th captures a third of the month. The realism factor exists to make projections honest about this rather than to make them look better.
Q6
Realized savings occasionally exceed projected. The usual cause is:
A. A pricing error in the rate card
B. Currency movement over the course of the whole billing month
C. Double-counting across two rules
D. Cascading savings: stopping one resource removes dependent charges
Show answer
Correct: D. Stopping an instance also stops its data transfer and its attached-volume IOPS. The rule priced the instance; the bill reflects the whole cascade.
Q7
Metrics sync writes into:
A. The recommendations table itself
B. A metric-points store the recommender reads over gRPC
C. A per-rule in-process cache
D. The cloud provider’s own native metric storage service
Show answer
Correct: B. Rules read from the store rather than calling cloud APIs at recommendation time, which is what makes the read path instant and the rules independently testable.
Q8
Per-provider metric lookback windows differ because:
A. Different rules each need a different lookback window length
B. Providers charge per metric query made
C. Customers configure the window per org
D. Each cloud retains metrics economically for a different period
Show answer
Correct: D. The window is sized to each cloud’s published retention so a single sync request never asks for data the API will silently drop. Rules declaring their own window are always within the provider default.
Q9
The Recent Activity tab is most useful for:
A. Making instance rightsizing decisions with confidence
B. Forecasting next-month spend
C. Confirming abandonment before deleting or terminating
D. Measuring tag coverage
Show answer
Correct: C. Zero utilization plus zero control-plane activity for 90 days is a much stronger case than either alone, and it is the evidence that survives a challenge from the owning team.
Q10
Activity sync does NOT capture:
A. In-guest activity such as SSH sessions or application traffic
B. Which principal created the resource
C. Control-plane operations performed against that resource
D. The timestamp of the last write to it
Show answer
Correct: A. It reads the cloud’s control plane. A resource nobody has reconfigured may still be serving production traffic, which is why activity evidence complements metrics rather than replacing them.
What’s next
Back to Reading evidence.