M2.9 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
Event Readiness differs from reactive autoscaling because it:
A. Pre-scales ahead of a known event rather than reacting to load
B. Scales considerably further than the autoscaler ever would
C. Uses different metrics to decide when
D. Works only on AWS scaling groups
Show answer
Correct: A. Reactive autoscaling responds after demand arrives, and the ramp-up gap is where a launch fails. Pre-scaling removes the gap for events you know about in advance.
Q2
Databases in an Event Readiness plan are:
A. Scaled the same way as compute targets are
B. Excluded entirely from the readiness plan
C. Attached as monitor-only targets with sizing recommendations
D. Scaled, but only after an explicit human approval is given
Show answer
Correct: C. The platform surfaces connection-pool math, headroom and a suggested tier, and never modifies the database. Tier changes and replica scaling stay with the DBA team.
Q3
The two capacity models are:
A. Linear and exponential growth
B. Multiplier and expected requests
C. Peak and average load models
D. Manual and automatic sizing
Show answer
Correct: B. Both reduce to the same equation (ZopNight turns expected requests into a multiplier by dividing by today’s request rate); they differ in which input you actually have. Per-target overrides exist for events where one component scales differently.
Q4
Scaling is executed via:
A. Continuous polling by ZopNight
B. Manual intervention at the event start
C. An always-on autoscaler policy
D. Cloud-native scheduled actions
Show answer
Correct: D. ASG scheduled actions, ECS Application Auto Scaling scheduled actions, Azure fixed-date profiles and GCP scaling schedules. No polling means no dependence on ZopNight being reachable at the event moment.
Q5
The event lifecycle runs:
A. draft, then active, then done
B. planned, then executing, then complete, and then finally closed and archived
C. draft → scheduling → scheduled → scaling_up → active → scaling_down → completed
D. created, then approved, then run
Show answer
Correct: C. With cancelled and failed as off-ramps, and concurrent schedule requests on the same event rejected so two operators cannot double-schedule it.
Q6
Rollback on completion or cancellation restores:
A. The autoscaler policy’s own configured default minimum values only
B. The original min, max and desired values recorded before the event
C. The cloud provider’s defaults for the type
D. Nothing; the added capacity simply stays
Show answer
Correct: B. Min and max are snapshotted from the policy when the event is created or edited; desired is recorded at schedule time, before the scale-up. The restore reads these recorded values, not the live cloud.
Q7
The cost estimate carries an isEstimated badge because:
A. The event has not run yet, so there is no measured history
B. Costs vary by region and cannot be fixed
C. Estimates are always approximate anyway
D. At least one rate came from fallback pricing, not live data
Show answer
Correct: D. The badge fires when a compute rate is not live pricing (even a bundled-table hit), a database rate falls back, or rows and totals drift. It has nothing to do with execution history.
Q8
The most important database signal before an event is:
A. Connection-pool math
B. CPU utilization headroom
C. Storage headroom on disk
D. Query latency percentiles
Show answer
Correct: A. Scaling the app tier multiplies connections. A database that is fine on CPU can exhaust connections the moment the app tier triples, and that is the failure that takes down the launch.
Q9
GCP event cancellation uses an explicit force-send on the schedules field because:
A. The API is slow on large maps
B. PATCH merges map keys, so deleting via PATCH would silently leave stale schedules
C. GCP requires a PUT rather than a PATCH for this particular scaling API call
D. It avoids provider rate limits
Show answer
Correct: B. A cancel that silently leaves the scaling schedule in place would scale for an event that is not happening, which is worse than failing loudly.
Q10
AWS ASG cancellation sweeps every matching scheduled action because:
A. Partial failures would otherwise leave orphaned actions
B. It is faster than deleting one by one
C. Scheduled actions simply cannot be deleted individually
D. The API only accepts bulk operations
Show answer
Correct: A. Deterministic schedule ids keyed on the event make re-queue a no-op, and the sweep makes cleanup complete even after a partial failure.
What’s next
Back to Event Readiness.