AI0-001 AI Infrastructure and Technologies Practice Question
An ML platform team is running a recommendation model on a Kubernetes cluster with GPU nodes. During peak traffic, inference pods are frequently evicted and restarted, causing latency spikes. The team wants to reduce restart frequency and keep GPU utilization high without changing the model. Which combination of Kubernetes configuration changes should they apply?
⚠ Common exam trap
The trap here is assuming that adding more replicas or an autoscaler prevents evictions, when eviction is governed by pod QoS class and priority rather than replica count.
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
✓
Set resource requests equal to limits for GPU and memory, and assign a high-priority PriorityClass to the inference pods
Evictions under node pressure are mitigated by raising pod QoS and priority. Setting requests equal to limits yields Guaranteed QoS, which protects the pods' resource allocation, and a high PriorityClass makes the kubelet evict other workloads first. Autoscalers and disruption budgets influence capacity and voluntary disruptions but do not shield the pods from pressure-driven eviction on a contended GPU node.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✓
Set resource requests equal to limits for GPU and memory, and assign a high-priority PriorityClass to the inference pods
Why this is correct
Setting requests equal to limits gives the pods Guaranteed QoS, which makes the kubelet far less likely to evict them under node pressure, and it reserves the GPU and memory they need. A high-priority PriorityClass ensures that if the node does come under pressure, the scheduler and kubelet prefer evicting lower-priority workloads instead of the inference pods, directly reducing restart frequency and stabilizing latency.
- ✗
Configure a PodDisruptionBudget with minAvailable set to zero and enable cluster autoscaler
Why it's wrong here
A PodDisruptionBudget with minAvailable of zero allows voluntary disruptions to take down all replicas, which increases rather than reduces the chance of losing serving capacity. Cluster autoscaler helps add nodes but does not protect running pods from eviction caused by resource pressure. Neither setting grants the inference pods priority over other workloads competing for GPU and memory on the same node.
- ✗
Use a Vertical Pod Autoscaler in recommendation mode and set the pod restart policy to Always
Why it's wrong here
Vertical Pod Autoscaler in recommendation mode only suggests resource values; it does not apply them or prevent eviction. Restart policy Always is the default for long-running pods and simply restarts containers after they terminate, so it does not stop the eviction events. This combination leaves the pods exposed to the same node pressure and does not raise their scheduling priority.
- ✗
Add a Horizontal Pod Autoscaler targeting CPU utilization and increase the pod replica count
Why it's wrong here
Scaling on CPU utilization is misleading for GPU inference workloads because CPU usage often stays low while the GPU is the actual bottleneck. Adding replicas increases capacity but does not prevent evictions, since eviction is driven by node resource pressure and pod priority, not replica count. This change also risks over-provisioning expensive GPU nodes without addressing the root cause of restarts.
About these practice questions
This AI0-001 question is part of Courseiva's 962-question bank — original exam-style content with full explanations and wrong-answer analysis, never real exam questions or exam dumps. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official CompTIA exam blueprint
This AI0-001 practice question is part of Courseiva's free CompTIA 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 AI0-001 exam.