M0.1 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
A bill row reads DataTransfer-Regional-Bytes, $612, for traffic between two AZs in one AWS region. The correct reading is:
A. Internet egress, mislabelled by the provider’s own billing code somewhere upstream
B. A billing error; intra-region traffic is free on AWS
C. Inter-region replication between two AWS regions
D. Cross-AZ traffic, charged at roughly $0.01/GB each direction on AWS, GCP and Azure alike
Show answer
Correct: D. Only traffic staying inside a single zone is free. “Cross-AZ is free on AWS” is the most common cost misreading in the curriculum, and a multi-zone Kubernetes cluster that was never pinned is the usual source.
Q2
The four line-item categories on any cloud invoice are:
A. Compute, storage, network and support charges only
B. Provisioned time, moved bytes, stored bytes, API requests
C. On-demand, reserved, spot and savings plan
D. Fixed, variable, amortized and unattributed
Show answer
Correct: B. The four categories are shapes of consumption, not product families. That framing is what lets you trace any dollar to the usage event that caused it, on any of the three clouds.
Q3
CloudWatch custom metrics are billed at:
A. Per API call rather than per metric stored
B. $0.10 per metric per month, with no free tier
C. $0.30 per metric per month for the first 10,000
D. Free up to 1,000 metrics, then $0.02 each
Show answer
Correct: C. One custom metric per pod across 200 pods is about $60/month in metric charges alone, before log ingestion at $0.50/GB and storage at $0.03/GB-month. The failure mode is a default retention of “Never expire” on a high-cardinality namespace.
Q4
You need to answer “which resource drove yesterday’s spike?” The right billing source characteristic is:
A. Highest granularity, accepting the latency that comes with it
B. Lowest latency, regardless of the granularity that is lost
C. The provider’s own console summary view
D. Whichever source the finance team already uses
Show answer
Correct: A. Granularity and timeliness trade against each other. A per-resource question needs a per-resource source, and every cloud bill is at least 24 hours behind real time, so the answer is never available instantly at that grain.
Q5
The minimum viable tag set is:
A. One tag: cost-centre, and nothing else
B. Whatever the cloud provider itself happens to mark as being required for it
C. Four tags: owner, environment, cost-centre and application (or the org’s equivalent)
D. Ten or more, to cover every reporting axis
Show answer
Correct: C. Four is the number that survives contact with engineers. The MVT exists because a tag policy nobody applies produces worse attribution than a small one everybody does.
Q6
Every cloud bill is at least 24 hours behind real time because:
A. Providers batch and issue all their invoices monthly, always well in arrears
B. Usage must be metered, rated and aggregated before it can be reported at resource grain
C. The billing exports run on a weekly cron
D. Currency conversion delays the publication
Show answer
Correct: B. This is why anomaly detection is a daily cron evaluating the previous full day rather than a live stream, and why an anomaly surfaces the day after the spend rather than the minute of it.
Q7
Mistake #8 on the ten-cost-mistakes list, cross-zone chatter, applies to:
A. GCP only
B. AWS only
C. AWS and Azure only
D. AWS, GCP and Azure
Show answer
Correct: D. All three charge inter-zone traffic in the same region. Detection is inter-zone egress consistently high; the fix is pinning workloads to a zone or using regional persistent disks.
Q8
The cost-attribution problem is that:
A. Tags are applied inconsistently, so spend cannot be traced to an owner
B. Cloud providers do not expose per-resource cost figures at all
C. Reserved instances hide the true effective rate
D. Currency conversion loses precision per row
Show answer
Correct: A. The bill knows what was spent; it does not know who is accountable. Everything in showback design exists to close that gap, and tag coverage is the measure of how far closed it is.
Q9
A bill row is NOT:
A. A record of usage that actually occurred
B. Attributable to a specific resource, always
C. Dated to the day the usage happened
D. Expressed in the billing currency
Show answer
Correct: B. Some line items never match a discovered resource: data transfer, taxes, fees, rows with no resource id, and services the platform does not discover. Those become unattributed spend, which is a permanent category rather than a tagging failure.
Q10
Over-retained logs and metrics appear on the ten-mistakes list because:
A. They slow down every query run against them
B. Cloud providers charge a penalty rate for retaining any of the data beyond a year
C. They break the org’s compliance requirements
D. Retention defaults are frequently “never expire”, so cost accrues indefinitely with no signal
Show answer
Correct: D. The absence of a retention policy is the defect. Nothing alerts on it because nothing is wrong; the bill simply grows, which is why it reaches a checklist rather than a monitor.
What’s next
Back to The cloud bill, decoded.