AZ-204 Develop Azure compute solutions Practice Test 2 — 15 Questions

Question 1

You are developing a serverless application using Azure Functions. The function needs to scale to zero when idle and handle high traffic spikes. Which THREE features should you implement?

Accepted Answer

Use a Consumption hosting plan. Option A (Use a Consumption hosting plan) is correct because the Consumption plan is the only Azure Functions hosting model that scales to zero when idle and bills per execution, which directly satisfies the scale-to-zero requirement. Option B (Use a queue trigger for processing) is correct because queue-triggered functions (e.g., Azure Storage Queue or Service Bus) decouple producers from consumers and let the scale controller add instances based on queue length, which is the standard pattern for absorbing high traffic spikes. Option C (Leverage the built-in scaling controller for automatic scaling) is correct because the Azure Functions scale controller monitors trigger metrics (queue depth, Event Hub events, etc.) and dynamically adds or removes worker instances, enabling elastic scaling under load. Option D (Enable the 'Always On' setting) is not correct because Always On keeps the app loaded and prevents idle unload, which contradicts the scale-to-zero requirement and is only available on App Service/Premium plans. Option E (Use a Premium hosting plan) is not correct because although Premium supports pre-warmed instances and VNet integration, it does not scale to zero when idle (it requires a minimum of one always-ready instance).

Answer

Enable the 'Always On' setting

Answer

Use a Premium hosting plan

Question 2

You are deploying a containerized application to Azure Container Instances (ACI). Which TWO actions should you take to ensure the container can securely access an Azure Key Vault?

Accepted Answer

Enable a managed identity for the container group. To securely access Azure Key Vault from Azure Container Instances, the recommended approach is to enable a managed identity for the container group. This allows the container to authenticate to Key Vault without requiring any credentials in the container image or code. The managed identity is integrated with Microsoft Entra ID, and the container can request an access token from the Azure Instance Metadata Service (IMDS) endpoint at 169.254.169.254. This token authorizes requests to Key Vault, ensuring secrets are never exposed. Note that Azure Container Instances does not support mounting a volume directly from Key Vault; the volume mount option shown in option C is not available in ACI.

Answer

Pass secrets as environment variables

Answer

Mount a volume that maps to a Key Vault secret store

Answer

Register the container with an Azure AD application

Answer

Configure a custom DNS server

Question 3

Refer to the exhibit. You are deploying this ARM template to create a deployment slot for an Azure App Service. The template deploys successfully. However, the slot-specific app setting 'DEPLOYMENT_SLOT' is not visible in the Azure portal under the slot's Configuration > Application settings. What is the most likely cause?

Accepted Answer

The template does not set 'slotSticky' to true for the app setting. In ARM templates, app settings are made slot-specific (sticky) by explicitly listing them in the 'appSettingNames' property of a 'Microsoft.Web/sites/slots/config' resource with the name 'slotConfigNames'. Without this configuration, the setting is treated as a regular, non-sticky app setting. While it will still appear under the slot's Configuration > Application settings in the Azure portal, it will not be marked as a 'Deployment slot setting' and thus will not be pinned to the deployment slot during a swap.

Answer

Slot settings are not supported in ARM templates

Answer

Deployment slots have been deprecated for App Service

Answer

The app setting name must be prefixed with 'APPSETTING_'

Question 4

Refer to the exhibit. You have an Azure App Service that uses the authentication settings shown. Users report that they are repeatedly prompted to sign in even after authenticating. What is the most likely cause?

Accepted Answer

The 'loginParameters' does not include 'offline_access'.. The most likely cause is that the 'loginParameters' does not include 'offline_access'. For Azure App Service Authentication (Easy Auth) to maintain a user's session without repeated sign-ins, it needs to be able to refresh the access token silently. This requires a refresh token, which is only issued by Microsoft Entra ID if the 'offline_access' scope is explicitly requested during the initial login. Without this scope, no refresh token is provided, and once the access token expires (typically after 1 hour), the user is prompted to sign in again.

Answer

The 'allowedAudiences' is missing the App Service URL.

Answer

The 'issuer' URL is incorrect; it should be a tenant-specific endpoint.

Answer

No 'tokenRefreshEndpoint' is configured for session management.

Question 5

Which TWO features of Azure App Service can help you reduce application downtime during deployments?

Accepted Answer

Deployment slots.. Deployment slots (C) let you deploy a new version to a separate staging slot with its own hostname while the production slot keeps serving traffic, so the app stays available until you are ready to promote the build. Slot swap with auto-swap (E) performs a VIP swap that warms up the target slot and swaps the hostnames, and auto-swap can trigger the swap automatically after a successful deployment, minimizing the cutover window and downtime. Continuous deployment from GitHub (A) only automates build and release triggering; it does not itself provide a zero-downtime cutover mechanism. Traffic Manager (B) is a DNS-based global load-balancing and failover service, not an App Service deployment feature that reduces downtime during a slot-based release. Auto-heal (D) restarts or recycles a misbehaving app based on rules, which addresses runtime faults rather than deployment downtime.

Answer

Continuous deployment from GitHub.

Answer

Traffic Manager.

Answer

Auto-heal.

Question 6

You are deploying an Azure App Service that must be accessible only from your corporate network via a VPN. You need to restrict inbound traffic. Which TWO actions should you take?

Accepted Answer

Deploy the App Service inside a VNet using an App Service Environment. Option B is correct because an App Service Environment (ASE) is a fully isolated, dedicated deployment of Azure App Service directly into a subnet of your VNet, which allows inbound access to be controlled entirely by the VNet and network security groups, so only traffic from the corporate VPN-connected network can reach it. Option E is correct because configuring IP address restrictions on the App Service lets you define an allow list of permitted client IP addresses or CIDR ranges, so you can allow only the corporate network's public IP addresses (for example, the VPN gateway's egress IPs) and deny everything else. Option A is not appropriate because Azure Front Door with WAF is a global edge service that exposes the app publicly and filters HTTP traffic, not a mechanism for restricting inbound access to a corporate VPN network. Option C is not correct because service endpoints secure access to Azure PaaS services from a VNet for outbound traffic, not inbound access to an App Service. Option D is not correct because VNet integration enables an App Service to make outbound calls into a VNet; it does not restrict inbound traffic to the app.

Answer

Use Azure Front Door with WAF

Answer

Configure a service endpoint

Answer

Enable VNet integration for the App Service

Question 7

You are developing a solution that processes orders. Each order must be processed exactly once, in the order it was received. You need to choose an Azure service that guarantees FIFO delivery and at-least-once processing. Which service should you use?

Accepted Answer

Azure Service Bus Queue with sessions. Azure Service Bus Queue with sessions is correct because it provides FIFO (first-in, first-out) delivery by grouping related messages into sessions, ensuring messages within a session are processed in order. The at-least-once processing guarantee is inherent to Service Bus queues, as messages are not removed until the receiver completes them, and they can be redelivered if the processing fails or times out.

Answer

Azure Event Hubs

Answer

Azure Event Grid

Answer

Azure Storage Queue

Question 8

You are developing a background job that runs every hour to process data from an Azure SQL database and send notifications via SendGrid. The job must be serverless and cost-effective, and must automatically retry on failure. Which TWO options meet the requirements?

Accepted Answer

Azure Functions with Timer trigger. Option B (Azure Functions with Timer trigger) is correct because Azure Functions is a fully serverless compute service that scales automatically and charges only for execution time, and its Timer trigger uses a CRON expression to run the job on an hourly schedule; Functions also provides built-in retry policies (e.g., configurable max retry count and exponential backoff via host.json) for automatic retries on failure. Option E (Azure Logic Apps with Recurrence trigger) is correct because Logic Apps is a serverless, consumption-based workflow service whose Recurrence trigger schedules hourly runs, and its SendGrid connector plus built-in retry policy (default exponential retry on HTTP 429/5xx) satisfies the notification and automatic-retry requirements without managing infrastructure. Option A (Azure Virtual Machine with a scheduled task) is not serverless — you pay for the VM continuously and must manage the OS and scheduler. Option C (Azure Batch with a schedule) is designed for large-scale parallel compute jobs and requires pool/node management, so it is not serverless or cost-effective for a simple hourly job. Option D (Azure WebJobs with TimerTrigger) runs inside an App Service plan (or App Service Environment) rather than a true serverless consumption model, so it does not meet the serverless requirement as cleanly as Functions or Logic Apps.

Answer

Azure Virtual Machine with a scheduled task

Answer

Azure Batch with a schedule

Answer

Azure WebJobs with TimerTrigger

Question 9

Your company has an Azure Kubernetes Service (AKS) cluster that hosts multiple microservices. You are tasked with deploying a new microservice that processes incoming HTTP requests and publishes messages to an Azure Service Bus topic. The microservice must scale based on the number of messages in the topic, and it must support graceful shutdown to complete in-flight requests. You need to choose the appropriate compute platform. The microservice is stateless and can be containerized. You want to minimize operational overhead and cost. The solution must automatically scale to zero when there are no messages. Which option should you choose?

Option A: Deploy the microservice as an Azure Function with a Service Bus trigger on the Consumption plan.
Option B: Deploy the microservice as a container in AKS with a Horizontal Pod Autoscaler based on Service Bus queue length.
Option C: Deploy the microservice as an Azure Container App with a Service Bus scale rule.
Option D: Deploy the microservice as an Azure App Service WebJob with continuous mode.

Accepted Answer

Deploy the microservice as an Azure Container App with a Service Bus scale rule.. Azure Container Apps (ACA) with a Service Bus scale rule is the correct choice because it provides event-driven scaling based on the number of messages in a Service Bus topic, can scale to zero when there are no messages, supports graceful shutdown via terminationGracePeriodSeconds, and minimizes operational overhead compared to AKS. ACA is a serverless container platform that abstracts Kubernetes complexity while still allowing containerized workloads, making it ideal for stateless microservices that need to scale on demand. Azure Functions with a Service Bus trigger also scales based on messages and can scale to zero, but it does not natively support containerization (unless using custom containers which adds complexity) and has less control over graceful shutdown. AKS requires managing a cluster and does not scale to zero. App Service WebJobs are not containerized and do not scale based on Service Bus metrics.

Answer

Deploy the microservice as an Azure Function with a Service Bus trigger on the Consumption plan.

Answer

AKS with HPA based on Service Bus queue length

Answer

Azure App Service WebJob with continuous mode

Question 10

You are developing a web application that allows users to upload images. The application is deployed on Azure App Service. After upload, the images must be processed to generate thumbnails and to extract metadata. The processing should happen asynchronously and must be resilient to failures. You need to design the solution using serverless components. The solution must minimize latency for the user during upload, and the processing must be retried automatically if it fails. You also need to ensure that the processing is idempotent, so that duplicate messages do not cause duplicate thumbnails. Which approach should you use?

Option A: Use Azure Functions with a Blob Storage trigger to process each image as it is uploaded. The function generates thumbnails and stores metadata in Cosmos DB. Use the `leaseBlob` property to prevent duplicate processing.
Option B: Use Azure Functions with an Event Grid trigger to process images. The function generates thumbnails and stores metadata in Cosmos DB. Use Event Grid's built-in retry policy and idempotent logic in the function.
Option C: Use Azure Logic Apps with a Blob Storage connector to process images. The logic app generates thumbnails and stores metadata in Cosmos DB. Configure retry policy on the connector.
Option D: Use Azure Functions with a Service Bus queue trigger. The web app sends a message to the queue after upload. The function processes the message, generates thumbnails, and stores metadata. Use message deduplication to ensure idempotency.

Accepted Answer

Azure Functions with Service Bus queue trigger and duplicate detection. It uses a Service Bus queue with duplicate detection, which ensures idempotent processing by automatically discarding duplicate messages within a defined time window. The web app uploads the image and immediately sends a message to the queue, minimizing user latency. The Azure Function triggered by the queue processes the image asynchronously, and Service Bus's built-in retry policy (via dead-lettering and max delivery count) provides resilience against failures.

Answer

Azure Functions with Blob Storage trigger, using leaseBlob

Answer

Azure Logic Apps with Blob Storage connector

Answer

Azure Functions with Event Grid trigger

Question 11

You are designing a serverless data processing pipeline. The pipeline receives JSON messages from an Azure Event Hubs instance. Each message must be enriched with data from a Cosmos DB database and then written to a Parquet file in Azure Data Lake Storage Gen2. The enrichment step involves a lookup that takes approximately 2 seconds per message. The pipeline must process up to 1000 messages per second. You need to choose the most cost-effective and scalable compute option. Consider the following options: A) Use a single Azure Function with Event Hubs trigger and output to Data Lake Storage. B) Use a Durable Functions orchestration with fan-out/fan-in pattern. C) Use Azure Stream Analytics with a reference data input from Cosmos DB and output to Data Lake Storage. D) Use an Azure Databricks notebook with structured streaming. Which option should you recommend?

Accepted Answer

Use Azure Stream Analytics with a reference data input from Cosmos DB and output to Data Lake Storage.. Azure Stream Analytics with a reference data input from Cosmos DB is the most cost-effective and scalable option because it can handle high-throughput streams (up to 1 GB/s) with sub-second latency, and it natively supports enriching incoming events with static or slowly-changing reference data (like Cosmos DB) without requiring custom code. The enrichment lookup is performed in-memory within the Stream Analytics job, avoiding per-message function invocation overhead and enabling linear scale-out across streaming units to meet 1000 messages/second with a 2-second lookup.

Answer

Use a single Azure Function with Event Hubs trigger and output to Data Lake Storage.

Answer

Use a Durable Functions orchestration with fan-out/fan-in pattern.

Answer

Use an Azure Databricks notebook with structured streaming.

Question 12

Your company is deploying a multi-container application using Azure Container Instances (ACI) in a virtual network. You need to ensure that containers can communicate with each other using localhost. Which TWO actions should you take?

Accepted Answer

Deploy all containers in the same container group.. Containers within the same container group in Azure Container Instances share the same network namespace, including the same IP address and port space. This allows them to communicate over localhost (127.0.0.1) without additional configuration, as they are essentially running on the same virtual machine. Option E is incorrect because using the container group's FQDN does not enable localhost communication; the FQDN resolves to the group's IP address, which is not localhost. Containers in the same group can communicate via localhost directly, without needing the FQDN.

Answer

Define environment variables on each container with the hostnames.

Answer

Assign different private IP addresses to each container.

Answer

Deploy containers in separate container groups and use service discovery.

Answer

Use the container group's fully qualified domain name (FQDN) to communicate.

Question 13

You are deploying an Azure App Service that uses a Linux container to host a custom web application. You need to configure continuous deployment from a GitHub repository. Which TWO actions should you take?

Accepted Answer

Use the App Service built-in CI/CD feature to connect to GitHub.. Azure App Service provides a built-in CI/CD feature that directly integrates with GitHub, enabling automatic deployment of code changes without additional configuration. Option C is also correct because the Deployment Center in App Service allows you to configure GitHub Actions as the CI/CD pipeline, which builds and deploys the container to the App Service on each push.

Answer

Set up an FTP trigger to poll the GitHub repository for changes.

Answer

Use the Kudu service to sync with GitHub.

Answer

Push the container image to Docker Hub and configure webhook.

Question 14

You are developing an Azure Functions app that processes events from an Event Hubs instance. The function must scale out automatically based on the number of partitions in the Event Hub. You need to ensure that each function instance processes events from at least one partition. Which TWO configurations should you use?

Accepted Answer

Set the function app to use the 'Event Scale' mode with a target of one instance per partition.. The 'Event Scale' mode with a target of one instance per partition ensures that the function app scales out to match the number of Event Hub partitions, with each instance processing events from at least one partition. Option C is correct because an event processor host with blob storage for checkpointing enables load balancing across multiple instances, ensuring each instance handles one or more partitions. Option E is incorrect because 'PartitionKey' is used when sending events to Event Hubs to assign a partition, not in the trigger binding; the trigger automatically distributes partitions across instances.

Answer

Set the 'MaxBatchSize' property to 1 to ensure even distribution.

Answer

Select the Premium App Service plan for the function app.

Answer

Use the EventHubs trigger with the 'PartitionKey' parameter set to the partition ID.

Question 15

You are deploying a critical web application on Azure App Service that must handle sudden traffic spikes. The application stores session data in memory. You need to ensure session state is preserved across scale-out events. Which THREE actions should you take?

Accepted Answer

Enable sticky sessions (ARR affinity) in the App Service configuration.. Option A is correct because enabling sticky sessions (ARR affinity) in App Service configuration ensures that a client's requests are routed to the same instance, preserving in-memory session state during scale-out. Option B is correct because Azure Cache for Redis provides a distributed, external session state store that all instances can access, eliminating the dependency on a single instance's memory. Option E is correct because a SQL Server session state provider also externalizes session data to a shared database, allowing any instance to retrieve the session regardless of which instance handles the request. Option C is incorrect because disabling ARR affinity would route requests to arbitrary instances, breaking in-memory session state. Option D is incorrect because 'InProc' session state stores data in the local process memory of a single instance, which is lost or inaccessible when scaling out.

Answer

Disable ARR affinity to allow any instance to handle requests.

Answer

Set the session state mode to 'InProc' in the web.config.