An ASG should maintain average CPU around 50% — scaling out when higher, in when lower — with minimal policy tuning. Which scaling policy type best fits?
A.Target tracking
Metric-target thermostat = target tracking. Correct.
B.Step scaling
Step scaling needs manual threshold/amount tuning — more control, more work.
C.Scheduled scaling
Scheduled handles predictable time patterns — not reactive CPU maintenance.
D.Predictive scaling
Predictive forecasts ahead — useful for known cyclical load, not the simple 'keep CPU at X' requirement.
Why this answer
Target tracking is 'set the thermostat': pick a metric target (CPU 50%, requests/target) and ASG manages scale-in/out continuously — simplest to configure. Step scaling = explicit tiered adjustments; scheduled = known time patterns; predictive = ML forecast-based pre-scaling.
Correct answer: A · Last reviewed 2026-11-02
Official AWS references
Practice this interactively
Answer without seeing the solution first in the High Availability practice session.
Practice High Availability