CLOUD-DIGITAL-LEADER practice question 9 of 35

Your organization wants to optimize its use of Google Cloud’s discounts on virtual machine-based workloads. You plan to use 200 CPUs constantly for the next 3 years, and you forecast that spikes of up to 300 CPUs will occur approximately 30% of the time. What should you choose?

Full set: 411 questions
Answer options
A: 1-year committed use discount for 200 CPUs
B: 3-year committed use discount for 300 CPUs
C: 3-year committed use discount for 200 CPUs Correct answer
D: Regular pay-as-you-go pricing
Correct Answer:
C. 3-year committed use discount for 200 CPUs
Explanation

A 200-vCPU baseline will run 100 % of the time for 3 years, so it is ideal for a 3-year committed-use discount (CUD), which gives the highest percentage reduction for long-term, predictable usage. Any usage above the 200-vCPU commitment (the extra 100 vCPUs needed about 30 % of the time) is simply billed at on-demand rates and still qualifies for sustained-use discounts when it runs long enough. Buying a larger (300 vCPU) commitment would waste money 70 % of the time, and a 1-year commitment leaves two years of higher pricing compared with a 3-year CUD.

Why Incorrect

A. 1-year CUD costs more overall than a 3-year CUD for the same constant workload.
B. Paying for 300 committed vCPUs overspends during the 70 % of time when only 200 are used.
D. On-demand pricing forfeits the large CUD savings for the always-on 200-vCPU baseline.

References

1. Google Cloud Documentation – “Committed use discounts,” Section “Choose your commitment amount” & “1-year vs 3-year discounts” (https://cloud.google.com/compute/docs/instances/signing-up-committed-use-discounts).

2. Google Cloud Documentation – “Compute Engine pricing,” Table “vCPU and memory committed-use discount rates” (accessed 2024-02-12).

3. Google Cloud Documentation – “Sustained use discounts,” Paragraph “How sustained use discounts work with committed use discounts.”

4. Google Cloud Architecture Framework – Cost Optimization: “Purchase CUDs for the steady-state portion of workloads; use on-demand for bursts” (v2023-10, Section 4.1).

Scroll to Top