AZ-305 Guide Path 5 · Module 1 · Design an Azure compute solution

Path 5 · Module 1

Design an Azure compute solution

📖 Microsoft Learn ~61 min❓ Quiz 20 questions✍ Design exercise 25 min

Read the module on Microsoft Learn, then come back here: test yourself with the quiz and apply it in the design exercise. Write down what you want to remember with the ✎ Note buttons.

Step 1 · Read the module

Open Design an Azure compute solution on Microsoft Learn. The units:

  1. Introduction
  2. Choose an Azure compute service
  3. Design for Azure Virtual Machines solutions
  4. Design for Azure Batch solutions
  5. Design for Azure App Service solutions
  6. Design for Azure Container Instances solutions
  7. Design for Azure Kubernetes Service solutions
  8. Design for Azure Functions solutions
  9. Design for Azure Logic Apps solutions
  10. Module assessment
  11. Summary and resources
Before you start the quiz

Close the Learn tab. Answering from memory is what makes it stick; looking it up just measures how fast you can scroll.

Part A

Quiz

20 exam-style questions, roughly in the order of the units. Click an answer to see the explanation. Your best score is saved and shown on the path page; use Retry at the end to go again.

1. Which question comes first in the compute decision tree?

A. Lift-and-shift points to IaaS; new or modernized apps open up PaaS, containers and serverless.

2. A legacy app needs full control of the operating system and custom software installs. Which service?

C. VMs give full OS control at the cost of managing patches and configuration.

3. A web tier must scale out automatically from 2 to 20 identical VMs based on CPU. What do you use?

C. Scale sets manage identical VMs and scale them with autoscale.

4. A VM workload needs a lot of memory relative to CPU (e.g. a large in-memory cache). Which VM family?

A. Memory-optimized sizes have a high memory-to-CPU ratio.

5. Which VM family suits a dev server that is mostly idle with occasional spikes, at low cost?

A. B-series VMs accumulate CPU credits when idle and burst when needed.

6. A rendering job runs thousands of independent tasks in parallel on hundreds of VMs, then stops. Which service?

C. Batch schedules large-scale parallel and HPC jobs on pools of VMs.

7. How can you reduce the cost of interruptible Batch or VM workloads?

D. Spot capacity is heavily discounted but can be evicted.

8. A .NET web app needs deployment slots, autoscale and custom domains, without managing servers. Which service?

A. App Service is the managed web hosting platform with slots and autoscale.

9. New code must be tested in production-like settings and swapped in with no downtime. Which App Service feature?

A. Slots let you warm up a new version and swap it into production.

10. An App Service app must reach a SQL database in a VNet. What do you configure?

C. VNet integration is for outbound traffic from the app into the VNet; private endpoints are for inbound.

11. A container must run a short task on demand, without any orchestration. Simplest option?

A. ACI runs containers quickly with per-second billing and no cluster to manage.

12. In ACI, which is the group of containers that share a lifecycle, network and storage on one host?

A. Container groups are similar to Kubernetes pods.

13. A microservices platform needs full Kubernetes APIs, custom operators and fine-grained network policies. Which service?

C. AKS gives full Kubernetes control; Container Apps abstracts Kubernetes away.

14. AKS workloads must scale pods based on CPU, and the cluster must add nodes when pods can't be scheduled. Which two features?

D. HPA scales pods; the cluster autoscaler scales nodes.

15. AKS needs some GPU nodes for ML and normal nodes for web apps. How do you design this?

A. Node pools let you mix VM sizes in one cluster; system pools run system pods.

16. Which Functions hosting plan scales to zero and bills per execution, but has cold starts?

A. Consumption plans are pay-per-use; Premium keeps pre-warmed instances and adds VNet features.

17. A function must avoid cold starts and access resources in a VNet. Which plan?

C. Premium provides pre-warmed instances and VNet integration.

18. A long-running, stateful workflow (fan-out/fan-in, waiting for human approval) must be written in code. What do you use?

C. Durable Functions add orchestrations with state, checkpoints and timers.

19. A business process must connect Office 365, Salesforce and SAP with a visual designer and little code. Which service?

A. Logic Apps is a low-code workflow service with hundreds of connectors.

20. Which Logic Apps plan runs in a single-tenant environment with VNet integration and built-in connectors running in-process?

B. Standard runs on the Functions runtime in your own plan; Consumption is multi-tenant pay-per-execution.

Part B

Design exercise

On paper · about 25 minutes

Quayside Ticketing sells concert tickets.

  1. Task

    Choose a compute service per workload

    Pick one service per workload and give the deciding requirement.

    Done when
    • Six workloads, six choices.
    • Each choice has a reason tied to the scenario.
    Model answer
    • Box office: VM (COM component, OS control).
    • Web shop: App Service with autoscale and deployment slots.
    • PDF tickets: Functions (queue trigger), Premium if cold starts hurt.
    • Seating maps: Azure Batch with Spot VMs.
    • Partner sync: Logic Apps.
    • Microservices: AKS.
  2. Task

    Design scaling and resilience for the web shop

    Describe how it survives the opening-sale peak and a zone failure.

    Done when
    • Autoscale rules are concrete.
    • Zone resilience is addressed.
    Model answer

    Premium v3 plan, zone-redundant (≥3 instances), autoscale on CPU/HTTP queue length with a scheduled scale-out before announced sales; static content via CDN/Front Door; purchases queued so back-end spikes don't fail the front end.

  3. Task

    Plan the AKS cluster

    Sketch node pools and autoscaling for the 25 microservices.

    Done when
    • System and user pools are separated.
    • Both pod and node scaling are covered.
    Model answer

    System node pool (3 nodes across zones), user node pool(s) for services, optional Spot pool for batch-like jobs; HPA per deployment, cluster autoscaler per pool; Entra ID integration, Azure CNI, managed identity, Container Insights.