Unit 4 Roadmap — Design infrastructure solutions
AZ-305 › Unit 4
Unit 4 Roadmap — infrastructure solutions
This roadmap turns the unit into a study sequence. The goal is to connect topic decisions, not memorize isolated product facts. The current exam blueprint allocates 30–35% to this domain.
Unit 4 release map
- 30–35%
- 4
- Defensible recommendation
- Checkpoint plus design studio
Unit thesis
Compose compute, integration, migration, and networking decisions into a secure, supportable target architecture.
Recommended learning sequence
- Select compute from execution model, control needs, scaling, state, and platform ownership.
- Separate messaging, events, APIs, configuration, and deployment concerns.
- Build migration waves and network paths from dependency and traffic evidence.
Text equivalent: study each topic cram sheet, complete the unit checkpoint, then prove synthesis in a design studio.
Topic map
| Topic | Architectural decision | First trap to reject |
|---|---|---|
| 1. Design compute solutions | Choose the least operationally heavy compute surface that still meets control, scaling, state, and runtime needs. | Containers do not automatically require AKS. |
| 2. Design an application architecture | Choose synchronous APIs, queues, events, workflows, caches, and configuration from coupling and delivery semantics. | Events and commands are not interchangeable. |
| 3. Design migrations | Use discovery evidence to choose a disposition, target landing zone, dependency-aware wave, and validated cutover. | A successful replication is not a successful migration. |
| 4. Design network solutions | Map communication flows to connectivity, routing, security, name resolution, load balancing, and observability controls. | NSGs do not provide application-layer inspection. |
How to work this unit
For each topic, read the linked full lessons first when the service boundary is unfamiliar. Use the cram sheet to compress the decision rules. Then close the notes and answer the retrieval prompts. Finish with scenarios that force two or more topics to interact. A correct product name without a constraint-based explanation is not sufficient evidence of readiness.
When reviewing an answer, write a one-sentence recommendation in this form: Choose X because constraints A and B matter; reject Y because it fails C; accept trade-off D. This structure exposes guesses and makes technical review easier.
Pre-checkpoint checklist
- I can state the decision boundary for every service family in this unit.
- I can distinguish high availability, recovery, security, and governance controls when they appear together.
- I can identify the strongest distractor and the requirement it fails.
- I can draw the major dependency or signal flow without copying a diagram.
- I can explain operational ownership and cost consequences, not only features.
- I have corrected every missed retrieval prompt at least once from memory.
Common failure patterns
- Choosing AKS when the requirements do not need Kubernetes control.
- Using a global routing service for private east-west connectivity.
- Migrating servers before identity, DNS, landing zone, and dependency readiness.
Source and freshness
Aligned to the current AZ-305 study guide and grounded in the two attached corpus sources. Product details and limits must be checked against current Microsoft Learn documentation. Reviewed 2026-08-02.