A reporting server will run an analytics engine that needs 8 vCPUs and 64 GiB of RAM. Average CPU use is expected to stay moderate, but the workload is memory heavy and should not use a burstable SKU. Which two deployment choices best align with the requirement? Select two.

Choose a burstable B-series size to lower cost.

Incorrect. Burstable SKUs are for light or variable workloads and usually have poor memory-to-CPU balance for analytics.

Pick a compute-optimized F-series size because CPU use is only moderate.

Incorrect. Compute-optimized sizes favor CPU-bound work, not memory-heavy analytics engines.

Select the smallest VM size that supports managed disks.

Incorrect. Disk compatibility does not address the stated CPU and memory needs of the workload.

What does this AZ-104 question test?

Read the scenario before looking for a memorised answer.

What is the correct answer to this question?

The correct answer is: Select a memory-optimized VM family. — The workload is explicitly memory heavy and needs a specific minimum of 8 vCPUs and 64 GiB of RAM. That means the administrator should start with a memory-optimized family rather than a burstable or compute-optimized series. The selected size must also satisfy the numeric resource target, because under-sizing memory can cause paging and poor query performance even when average CPU looks acceptable. Why others are wrong: Burstable sizes are a poor fit for sustained analytics and can throttle under load. Compute-optimized families are tuned for CPU-intensive workloads, which is not the main bottleneck here. Disk support alone does not make a VM suitable; the memory and CPU profile must match the application first.

What should I do if I get this AZ-104 question wrong?

Then try more questions from the same exam bank and focus on understanding why the wrong options are tempting.