Microsoft AI-200 Exam Prep
Developing AI Cloud Solutions on Azure (Page 2 )

Updated On: 31-Aug-2026
View Related Case Study

HOTSPOT (Drag and Drop is not supported)
You need to configure the database resources for the Azure Database for PostgreSQL instance.
How should you complete the configuration to meet the business and technical requirements? To answer, select the appropriate options in the answer area.
Note: Each correct selection is worth one point.
Hot Area:

  1. See Explanation section for answer.

Answer(s): A

Explanation:




Box 1: Increase compute vCores Technical requirements, Performance: Semantic search latency must remain under 200 milliseconds at peak load.
To reduce semantic search latency under 200 milliseconds at peak load, the best action is to Increase compute vCores.
Vector similarity search and semantic retrieval are highly CPU-intensive operations. The math behind vector distance calculations (such as Cosine similarity, Dot Product, or Euclidean distance) relies heavily on mathematical operations executed per query.
When a system experiences peak load, compute vCores easily become the primary bottleneck. Adding more vCores directly increases parallel processing capacity, dramatically shortening the execution time of similarity calculations and keeping retrieval latency within the required 200 ms SLA.
Box 2: Increase memory allocation The best action is to increase memory allocation.
High-dimensional vector indexes (such as HNSW or IVFFlat managed by the pgvector extension) are highly resource-intensive and rely heavily on RAM. To maintain fast vector similarity searches and avoid high-latency disk operations, the entire vector index must fit into memory (RAM residency). Scaling up the database instance's memory directly expands the PostgreSQL shared buffers and cache, ensuring the high-dimensional index remains resident in RAM for rapid semantic retrieval.
Box 3: Enable storage autoscale Enable storage autoscale is the best action to support the continuous ingestion of transaction-based embeddings.
Continuous Ingestion Demands Dynamic Space: Continuous transaction processing causes vector databases (such as Azure Database for PostgreSQL with pgvector or Azure SQL Database) to expand constantly over time.
Preventing Ingestion Failures: If storage reaches capacity limits, the database switches into a read-only state. This immediately fails and halts all incoming real-time embedding write operations. Enabling storage autoscale allows the environment to dynamically provision storage on the fly without downtime.


Reference:

https://dl.acm.org/doi/10.1145/3695053.3731013 https://learn.microsoft.com/en-us/training/paths/develop-ai-solutions-azure-database-postgresql/ https://learn.microsoft.com/en-us/azure/architecture/guide/technology-choices/vector-search



View Related Case Study

HOTSPOT (Drag and Drop is not supported)
You need to deploy a batch retraining workload.
How should you complete the scaling configuration? To answer, select the appropriate options in the answer area.
Note: Each correct selection is worth one point.
Hot Area:

  1. See Explanation section for answer.

Answer(s): A

Explanation:




Box 1: Azure Service Bus scaler An Azure Service Bus scaler needs to be set up to meet this requirement.
Queue-Depth Metrics: To trigger scaling based strictly on queue depth, the orchestration layer (such as Azure Container Apps or AKS using KEDA) requires a scaler that can natively communicate with the message broker and monitor metrics like active message count.
Scale-to-Zero Support: Unlike standard resource scalers, an Azure Service Bus scaler enables event-driven batch workloads to spin up consumers when messages arrive and scale all the way down to zero instances when the queue is completely empty.
Box 2: Minimum replicas = 0. You need to set Minimum replicas = 0.
Setting the minimum replicas to 0 enables the system to scale down to zero instances when there are no messages left in the queue, completely deallocating resources and eliminating idle compute costs.
Box 3: Maximum replicas = 10 To prevent uncontrolled burst scaling during your batch retraining workloads, you need to set maximum replicas = 10.
Enforcing Upper Bounds: Allowing the system to determine the maximum replica gives the underlying autoscaler (such as KEDA or the Horizontal Pod Autoscaler in Azure Kubernetes Service) the freedom to scale out infinitely or up to large default platform limits. This directly risks uncontrolled burst scaling when a massive batch queue is processed.
Resource and Cost Protection: Hard-coding a ceiling (like maximum replicas = 10) ensures that the batch workload cannot consume more cluster resources than allocated, preventing resource starvation for customer-facing recommendation APIs and keeping operational costs predictable.


Reference:

https://azure.github.io/aca-dotnet-workshop/aca/09-aca-autoscale-keda/ https://kserve.github.io/website/docs/reference/crd-api



You maintain multiple versions of a container image in Azure Container Registry.
The production deployment must always run the exact same image build even if tags are changed later.
You need to ensure predictable and immutable image selection during deployment.
What should you do?

  1. Tag the image as production and deploy it by using the production tag.
  2. Schedule nightly rebuilds of the image.
  3. Configure deployment to use the latest tag.
  4. Identify the image by using its SHA digest.

Answer(s): D

Explanation:

Deploying a container image by its unique SHA-256 digest digest instead of its tag ensures a predictable and immutable deployment. Tags can be overwritten or moved to different image builds, but a digest is a cryptographic hash that never changes.


Reference:

https://container-registry.com/posts/container-image-versioning/



A container in an AKS cluster repeatedly restarts.
Pod events show probe failures, although node-level CPU and memory metrics are normal.
You need to diagnose the cause of the repeating restarts.
What should you do first?

  1. Scale the deployment to more replicas.
  2. Decrease the initialDelaySeconds for the container liveness probe.
  3. Drain and reboot the node hosting the pod.
  4. Inspect the pod events and container logs.

Answer(s): D

Explanation:

Inspecting the pod events and container logs is the correct first action to take.
Because node-level metrics are normal, the issue is isolated to the specific workload or its individual container boundaries. Pod events will clarify exactly which health probe (liveness or readiness) is triggering the restart, and the logs will reveal why the application inside the container is failing to respond.


Reference:

https://aws.plainenglish.io/why-your-kubernetes-pod-is-crashing-and-how-to-fix-it-5c71f8d2779a? gi=a46b335270ff



You develop a message-processing service deployed to Azure Container Apps. The service reads messages from an Azure Service Bus queue.
The solution must minimize costs by ensuring NO compute resources are consumed when the queue is empty.
You need to configure scaling for the service.
Which two actions should you perform? Each correct answer presents part of the solution.
Note: Each correct selection is worth one point.

  1. Increase the scaling rule to allow for the maximum running replica count.
  2. Configure the scaling rule to allow for the termination of all active replicas.
  3. Configure a Kubernetes Event-driven Autoscaler rule that monitors queue length.
  4. Enable HTTP ingress concurrency scaling.

Answer(s): B,C

Explanation:

[C] You can configure an Azure Service Bus scale rule in Azure Container Apps, which utilizes built-in KEDA (Kubernetes Event-driven Autoscaling) to monitor queue length and scale the app from zero instances to meet your processing demands.
[B] Configure the scaling rule to allow for the termination of all active replicas is the correct action to take.
To ensure that no compute resources are consumed when the queue is empty, you must set the minimum replica count (minReplicas) to 0 in your Azure Container Apps scaling configuration.
When the Azure Service Bus queue has zero messages, KEDA will scale the container replicas down to zero, stopping all compute billing.


Reference:

https://learn.microsoft.com/en-us/azure/container-apps/scale-app



Viewing page 2 of 27
Viewing questions 6 - 10 out of 140 questions


Post your Comments and Discuss Microsoft AI-200 exam prep with other Community members:

AI Tutor AI Tutor 👋 I’m here to help!