When designing solutions that scale elastically, which of the following practices is most effective for both performance and cost efficiency?

Question

Options

• AAlways use the maximum available compute capacity for each instance
• BImplement auto-scaling based on real-time demand
• CDeploy services exclusively in a single geographic location to reduce complexity
• DUse reserved instances for all workloads to reduce costs

• A
  4% (2)
• B
  85% (39)
• C
  2% (1)
• D
  9% (4)

Explanation

Implementing auto-scaling based on real-time demand is the cornerstone of elastic architecture because it automatically adjusts compute resources to match actual load - adding capacity under peak demand and releasing it during quiet periods, simultaneously optimizing performance and minimizing cost. Option A wastes money by running at maximum capacity even when demand is low, paying for resources you don't need. Option C introduces a single point of failure and ignores latency benefits of geographic distribution - the opposite of scalable design. Option D uses reserved instances for all workloads, which suits predictable, steady-state workloads but is wasteful for variable or unpredictable traffic patterns that benefit more from on-demand pricing.

Memory tip: Think of auto-scaling like a thermostat - it continuously senses conditions and adjusts output to match need, rather than blasting heat all day (A), huddling in one room (C), or pre-paying for a fixed temperature year-round (D).

No community discussion yet for this question.