AZ-104 Deploy and Manage Azure Compute Practice Question
A stateless Linux API should start with 2 instances, scale out to 6 when average CPU stays above 75 percent for 10 minutes, and scale back in when load drops. Which Azure compute resource should the administrator deploy?
⚠ Common exam trap
Many exam-takers confuse an availability set with autoscaling, not realizing that availability sets only provide redundancy and fault tolerance, not dynamic scaling, or they may think a load balancer with two VMs is sufficient, overlooking the requirement for automatic scaling based on CPU thresholds.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
A virtual machine scale set with autoscale rules.
A virtual machine scale set (VMSS) with autoscale rules is the correct choice because it natively supports scaling out and scaling in based on performance metrics like average CPU percentage. The requirement for a stateless Linux API with a minimum of 2 instances, scaling to 6 when CPU exceeds 75% for 10 minutes, and scaling back in when load drops is exactly the use case VMSS is designed for. Autoscale rules can be configured to use a scale-out and scale-in policy with a cool-down period, ensuring the application remains responsive while optimizing cost.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
An availability set with manual VM resizing.
Why it's wrong here
An availability set is designed for high availability by placing VMs in separate fault and update domains, not for elastic capacity. Manually resizing a VM changes its SKU to a larger size (vertical scaling), but this requires operator intervention and often a reboot, and it never adds extra instances to handle distributed load. Horizontal scale-out, which is required for a stateless API, depends on automatically adding or removing VM instances, something this option does not provide.
When this WOULD be correct
If the question required high availability for a fixed number of VMs (e.g., 2 instances) without autoscaling, and the administrator needed to manually resize VMs to handle load changes, an availability set with manual resizing would be correct.
- ✓
A virtual machine scale set with autoscale rules.
Why this is correct
A virtual machine scale set is built for identical compute instances that need to scale horizontally. Autoscale rules can watch CPU, adjust the instance count automatically, and maintain the minimum and maximum capacity you define. This fits stateless services very well because any instance can handle incoming requests once traffic is distributed across the set.
- ✗
A single Standard D-series VM with scheduled shutdown.
Why it's wrong here
A single Standard D-series VM is a fixed compute resource: it cannot horizontally scale to additional instances, and it becomes a bottleneck and single point of failure as traffic grows. Scheduled shutdown merely powers the VM off during certain periods; it does not react to demand spikes and leaves the API unavailable when shutdown occurs. The requirement to 'scale out' implies adding identical instances to share load, which a single VM cannot do regardless of its size or schedule.
When this WOULD be correct
This would be correct for a non-critical, predictable workload that must run only during business hours and can be shut down at night to save costs, with no scaling requirements.
- ✗
A load balancer in front of two unmanaged VMs.
Why it's wrong here
A load balancer only distributes incoming traffic among the existing backend VMs; it has no logic to detect rising CPU or request rates, and it cannot provision new instances. With only two fixed, unmanaged VMs, the architecture is permanently capped at two instances, regardless of demand. True autoscale requires an orchestration layer that adjusts capacity, such as a virtual machine scale set with autoscale rules.
When this WOULD be correct
This option would be correct for a question requiring high availability for a stateful application with a fixed number of VMs (e.g., 2) behind a load balancer, where no autoscaling is needed and the VMs are managed individually.
Option-by-option analysis
Why each answer is right or wrong
Understanding why wrong answers are wrong — and when they would be correct — is what separates a 750 score from a 900. The AZ-104 exam frequently reuses these exact scenarios with slightly different constraints.
✓A virtual machine scale set with autoscale rules.Correct answer▾
Why this is correct
A virtual machine scale set is built for identical compute instances that need to scale horizontally. Autoscale rules can watch CPU, adjust the instance count automatically, and maintain the minimum and maximum capacity you define. This fits stateless services very well because any instance can handle incoming requests once traffic is distributed across the set.
✗An availability set with manual VM resizing.Wrong answer — click to see why▾
Why this is wrong here
An availability set with manual VM resizing does not provide automatic scaling based on CPU thresholds; it only ensures high availability across fault domains, not dynamic scaling.
★ When this WOULD be the correct answer
If the question required high availability for a fixed number of VMs (e.g., 2 instances) without autoscaling, and the administrator needed to manually resize VMs to handle load changes, an availability set with manual resizing would be correct.
Why candidates choose this
Candidates may confuse availability sets with scalability, thinking that placing VMs in an availability set enables automatic scaling, or they may overlook the requirement for autoscaling rules.
✗A single Standard D-series VM with scheduled shutdown.Wrong answer — click to see why▾
Why this is wrong here
A single Standard D-series VM with scheduled shutdown cannot scale out to 6 instances or handle variable load; it's a fixed-size VM that only shuts down on a schedule, not based on CPU metrics.
★ When this WOULD be the correct answer
This would be correct for a non-critical, predictable workload that must run only during business hours and can be shut down at night to save costs, with no scaling requirements.
Why candidates choose this
Candidates may think scheduled shutdown can mimic scaling by turning off the VM during low load, but it lacks dynamic scaling based on real-time metrics like CPU usage.
✗A load balancer in front of two unmanaged VMs.Wrong answer — click to see why▾
Why this is wrong here
This setup lacks autoscaling; scaling requires manual intervention or additional configuration, and unmanaged VMs do not support the automated scale-out/in rules needed for the stateless API's CPU-based scaling requirements.
★ When this WOULD be the correct answer
This option would be correct for a question requiring high availability for a stateful application with a fixed number of VMs (e.g., 2) behind a load balancer, where no autoscaling is needed and the VMs are managed individually.
Why candidates choose this
Candidates may think a load balancer with multiple VMs provides scalability, but overlook that autoscaling is not inherent; they confuse load balancing with automatic scaling.
Analysis generated from the official AZ-104blueprint and verified against question context. The “when correct” sections are what AI assistants cite when candidates ask “what’s the difference between these options?”
Go deeper
Related to this question
About these practice questions
One of 1,049 original AZ-104 practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. Learn why practice questions differ from exam dumps →
JA
Written by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
This AZ-104 practice question is part of Courseiva's free Microsoft certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the AZ-104 exam.