Azure AZ-305 prep, solutions architect expert roadmap with ARIA
The Microsoft Azure Solutions Architect Expert (AZ-305) is a 120-minute exam with around 40 questions including case studies, a 700 of 1000 passing score, and expert-level difficulty. Microsoft strongly recommends AZ-104 and AZ-204 first, since AZ-305 expects associate-level fluency across administration and development before it even starts. I prep you for it with an adaptive evaluation, a longer roadmap weighted toward design-of-infrastructure decisions, scenario and case-study drilling, and a pass guarantee tied to five measurable conditions. Start your free CAT evaluation at claudelab.me/onboarding/select-cert?code=AZ-305.
TL;DR
- 120 minutes, around 40 questions including case studies, 700 of 1000 to pass, expert level.
- Four design domains, with Design Infrastructure the heaviest at 35 percent and Identity and Security plus Data Storage tied at 25 percent.
- AZ-104 and AZ-204 are strongly recommended first; AZ-305 assumes associate-level Azure fluency from the first question.
- Roadmaps for AZ-305 typically run 8 to 12 weeks because the exam tests design judgement, not service recall.
- Pass-guarantee eligibility is checked by a database function with five mechanical conditions, not a marketing line.
What the AZ-305 exam is
AZ-305 is the current Microsoft Azure Solutions Architect Expert exam (current as of 2026). It tests your ability to design solutions on Azure across identity, data, business continuity, and infrastructure. Around 40 questions in 120 minutes, scaled passing score 700 out of 1000, with multiple choice, multiple response, drag-and-drop ordering, hot-area diagrams, and case studies that bundle four to six questions under a shared multi-page scenario.
The blueprint splits into four design domains:
| Domain | Weight | What it covers |
|---|---|---|
| Design Identity, Governance, and Monitoring Solutions | 25% | Microsoft Entra ID, External ID, B2C, conditional access, RBAC at scale, management groups, Azure Policy and Initiatives, Log Analytics design, cost management. |
| Design Data Storage Solutions | 25% | Cosmos DB consistency models and partitioning, Azure SQL deployment options, storage account design, blob lifecycle, data lake layout, encryption and key management. |
| Design Business Continuity Solutions | 15% | HA vs DR, zone-redundant vs region-pair, Azure Site Recovery vs Azure Backup, RPO and RTO sizing, multi-region application patterns. |
| Design Infrastructure Solutions | 35% | Compute selection (AKS, App Service, Container Apps, Functions, VMs), network topology (hub-and-spoke vs Virtual WAN), Front Door vs Application Gateway, integration and migration design. |
Design Infrastructure at 35 percent is the largest domain and the most common reason candidates fail this exam. A roadmap that spends equal time on all four wastes about a quarter of your study window. I do not.
The case-study format separates AZ-305 from associate-tier exams. A case study opens with a long scenario covering business goals, technical requirements, the existing environment, and planned changes, then asks four to six related questions against shared context. Misread the requirements once and the whole block bleeds.
Positioning vs SAP-C02 matters if you architect on both clouds. Both are expert-level architecture exams, both reward design judgement, both lean on case studies. AZ-305 quotes Microsoft Entra, Cosmos DB, AKS, Virtual WAN, and Azure governance. SAP-C02 quotes AWS Organizations, Control Tower, Transit Gateway, and a different service surface. Cross-cloud skills transfer, the exam stems do not. Recommended prereq for AZ-305: the AZ-104 administrator page.
How ARIA preps you for it
ARIA owns your AZ-305 prep end to end. Four pieces, all tuned for an expert-level exam.
The CAT evaluation, with design-thinking probes. Your first session is a 15-to-25-question adaptive test that converges on your skill level for each of the four AZ-305 domains. The difference from associate-tier evaluations: I deliberately mix service-knowledge probes with design-thinking probes, because AZ-305 is failed more often on design judgement than on whether you remember a service exists. A candidate who can name every Azure data service but cannot pick the right consistency model under a latency constraint will fail. The CAT surfaces both gaps separately. Read the full CAT explainer for the mechanics.
The longer roadmap, weighted toward Design Infrastructure. Expert exams need more milestones, not fewer. AZ-305 roadmaps typically run 8 to 12 weeks at 30 to 45 minutes a day, with Design Infrastructure (35 percent) and the two 25-percent design domains soaking up most of the milestone count. A novice on Design Infrastructure gets the most milestones in the plan. Full structure: the roadmap overview.
The gauntlet, with case-study emphasis. The gauntlet is a long-form session under exam conditions, unlocked at 80 readiness. For AZ-305, gauntlets weight case studies higher than the standalone-question mix in shorter practice, because case studies are the format most candidates have least practice on and the format that punishes pattern-only studying. See the gauntlet page and practice sessions.
The error backlog with architecture-trap categorization. Every wrong answer is tagged with a specific architecture trap: container-platform selection, network topology, consistency-model selection, HA vs DR boundary, identity-product selection, governance-construct overlap, cost-design tradeoff. Tags are diagnostic, not generic. The backlog reschedules misses at 1, 3, 7, and 21 days. A pattern retires only after three correct answers in a row, spaced. Background reading: AI cert prep in 2026. For the readiness math, see readiness and decay, and for eligibility logic, the pass guarantee page.
Common pitfalls on AZ-305
These are the seven traps that quietly cost the most points on this exam. Each one is a design-judgement failure, not a service-recall failure. That is the whole shape of AZ-305.
1. AKS vs Container Apps vs Container Instances
The trap: three container platforms, three sweet spots. AKS is full Kubernetes for teams that need cluster-level control and operational depth. Container Apps is managed Kubernetes-on-rails with built-in scale-to-zero, Dapr, and KEDA-driven autoscaling, intended for microservices that should not require a cluster operator. Container Instances is single-container or pod-level, no orchestration, billed per second, intended for short-lived jobs or sidecar workloads. The exam writes stems where two answers look right until you read the operational constraints (team capability, scale-to-zero need, multi-service orchestration) carefully.
What I do: every miss tags the container-platform selection trap, and the backlog rotates the three-way comparison until you stop reaching for AKS when Container Apps would do the job at half the operational cost.
2. Hub-and-spoke vs Virtual WAN
The trap: both are valid Azure network topologies. Hub-and-spoke is the classic pattern, one central VNet hosting shared services, spokes peered in, transit handled by a network virtual appliance or Azure Firewall in the hub. Virtual WAN is Microsoft's managed any-to-any backbone, with VPN, ExpressRoute, and SD-WAN integration, intended for organizations whose footprint outgrows a hand-managed peering mesh. Hub-and-spoke wins on simplicity at small scale; Virtual WAN wins on scale, branch connectivity, and operational overhead reduction. The exam stems hide the deciding factor in the requirements paragraph (branch count, transit needs, growth trajectory).
What I do: I drill the topology-decision trap with paired scenarios where the only thing that changes is one constraint, and you have to call which topology that constraint flips.
3. Cosmos DB consistency model selection
The trap: Cosmos DB offers five consistency levels, Strong, Bounded Staleness, Session, Consistent Prefix, and Eventual. Strong gives linearizability across regions but kills latency and forces single-region writes for some configurations. Bounded Staleness gives a configurable lag window. Session is the per-client guarantee that powers most user-facing apps. Consistent Prefix and Eventual trade more consistency for more throughput. The exam writes stems where the requirement is "lowest possible latency with read-your-own-writes" (Session, not Strong) or "globally consistent counters" (Strong, with the latency price you accept for it).
What I do: every miss tags the consistency-model trap, and the backlog drills the latency / throughput / consistency triangle until you stop defaulting to Strong out of caution.
4. HA vs DR design (zone-redundant vs region-pair, ASR vs Backup)
The trap: high availability and disaster recovery are not the same problem. HA is keeping a service up through component failure inside a region (availability zones, zone-redundant services, load balancing). DR is surviving the loss of an entire region (region pairs, Azure Site Recovery for replication and failover, geo-redundant storage). Azure Backup is point-in-time recovery, not failover. The exam writes stems where the requirement is "RPO of 15 minutes and failover to a paired region" (ASR plus zone-redundant primary, not just Backup) or "restore yesterday's deleted blob" (Backup, not ASR).
What I do: I split HA and DR into separate sub-patterns, then drill the boundary scenarios. The backlog brings back paired RPO / RTO stems until you can map a requirement to the right tool on first read.
5. Identity design (Entra ID vs External ID vs B2C, conditional access)
The trap: Microsoft Entra ID is the directory for your workforce. External ID (the rebrand of External Identities, replacing the standalone B2C product going forward) covers business-to-business and customer scenarios. Older AZ-305 study material still references Azure AD B2C as a separate tenant model. Conditional access policies sit on top, gating sign-in by signal (location, device compliance, risk score). The exam writes stems where the customer-vs-workforce distinction picks the product, and the policy-design distinction picks the access shape.
What I do: I tag identity-product selection separately from conditional-access design, because they fail for different reasons. The backlog rotates the use-case shapes until you can call the right product from the customer description alone.
6. Governance at scale (Management Groups + Policy + Initiatives vs Blueprints)
The trap: Azure governance has overlapping constructs that look interchangeable until you read the lifecycle and assignment scope. Management groups stack subscriptions for billing and policy hierarchy. Azure Policy enforces individual rules. Initiatives bundle policies under a single assignment. Blueprints bundle policies, RBAC assignments, ARM templates, and resource groups into a deployable artifact (with a deprecation trajectory you should know about). The exam writes stems where any of three answers look right and only one matches the actual requirement (rule enforcement vs deployable scaffolding vs hierarchical scope).
What I do: I tag each as a separate sub-pattern, drill the lifecycle distinction (policy is enforced continuously, blueprints are deployed), and rotate them in the backlog until you can read the requirement and call the correct construct without hesitation.
7. Cost design tradeoffs at the architecture level
The trap: AZ-305 expects you to factor cost into the design, not just to know what each service costs. Reservations are 1-or-3-year commitments that trade flexibility for steep discounts on steady workloads. Spot VMs trade availability for very deep discounts on interruptible work. Azure Savings Plans give a more flexible commitment across compute services for less discount than reservations. The exam writes stems where the workload shape (steady vs bursty vs interruptible) decides the answer, and the wrong choice costs the customer money the question quietly states they cannot spend.
What I do: every miss tags the cost-design trap, and the backlog brings back workload-shape scenarios until you stop reaching for reservations when savings plans fit better.
Common questions
Do I need AZ-104 and AZ-204 before sitting AZ-305?
Microsoft strongly recommends associate-level Azure experience, which is what AZ-104 (administrator) and AZ-204 (developer) cover together. Neither is a hard prerequisite, but AZ-305 assumes you can reason about VNets, identity, storage redundancy, and app platform tradeoffs without first being taught what each service does. If you are coming in cold, start at AZ-104 and add AZ-204 if your role touches code.
How are case-study questions scored on AZ-305?
Case studies present a long multi-page scenario (business goals, technical requirements, existing environment, planned changes) followed by clusters of four to six questions against shared context. Each question is scored individually, but they share the scenario, so a misread early in the case quietly costs you points later. AZ-305 leans heavier on case studies than AZ-104. I drill the read-order so you parse requirements, constraints, and existing state before opening question one.
How long does AZ-305 prep take at 30, 45, or 60 minutes a day?
At 30 minutes a day, plan on 12 to 16 weeks. At 45 minutes a day, 9 to 12 weeks. At 60 minutes a day, 8 to 10 weeks. Your CAT baseline matters more than raw hours: a candidate fresh off AZ-104 plus AZ-204 finishes faster than someone whose Azure background is uneven across compute, identity, and data.
Is AZ-305 multiple choice or does it have a hands-on lab?
AZ-305 has no graded hands-on lab. The format is multiple choice, multiple response, drag-and-drop ordering, hot-area diagrams, and case studies. It tests architecture decisions, not console operation. Prior hands-on Azure experience still helps you read stems faster, because architecture questions assume you already know what each service does and what its limits are.
Should I take AZ-305 or AWS SAP-C02 if I already know one cloud?
Take the one your work uses. Both are expert-level architecture exams, both reward design judgement, both lean on case studies. AZ-305 quotes Microsoft Entra, Cosmos DB, AKS, Virtual WAN, and Azure governance constructs. SAP-C02 quotes AWS Organizations, Control Tower, Transit Gateway, and a different service surface. Skills transfer, the exam stems do not.
Does the pass guarantee really cover AZ-305?
Yes, with the same five measurable conditions used on every cert: every milestone completed, every phase completed, two mock exams passed at the certification's passing score (Microsoft scales AZ-305 to 700 of 1000), one gauntlet passed at 80 percent or higher, and a live readiness score of 80 or above. Sit the exam in the 60-day window, fail, get a full refund of the Exam Ready plan. Full mechanics: the pass guarantee page.
Start your AZ-305 prep
The cheapest possible signal is the 15-minute CAT evaluation. It tells you which of the four AZ-305 design domains you actually own, which one will cost you the exam if you sit it tomorrow, and where the roadmap starts. AZ-305 is an expert exam, so the honest answer the CAT gives you is more useful than another month of unmeasured study.
Start your free AZ-305 evaluation at claudelab.me/onboarding/select-cert?code=AZ-305.