Examworthyexamworthy.comDesigning Microsoft Azure Infrastructure Solutions (AZ-305) cheat sheet
Microsoft
Free to share. Examworthy is not affiliated with or endorsed by Microsoft; AZ-305 and related marks belong to their respective owners.
At a glance
Format: Multiple choice, multiple response, and case studies, at a Pearson VUE testing center or online proctored
Domain weight map
Heaviest first - spend your time hereHow this exam thinks
AZ-305 is a requirement-to-design exam: almost every question is a scenario with constraints on identity, governance, recovery objectives, cost, and operations, and the right answer is the Azure service, tier, or topology that satisfies every constraint, usually the managed option that meets the requirement at the least overhead.
Spot the trap
Tempting wrong answers, and why they failTempting but wrong
The legacy Azure Functions Consumption plan allows unbounded execution up to 60 minutes per invocation.
Why it fails
Unbounded execution is documented for Flex Consumption, Premium, Dedicated, and Container Apps - not the legacy Consumption plan. Legacy Consumption is capped at a 10-minute maximum timeout with a 5-minute default, so a job longer than 10 minutes will not fit.
Design Infrastructure Solutions
Tempting but wrong
For a Foundry-hosted AI agent, you should wire Foundry traces straight to a Log Analytics workspace and skip Application Insights.
Why it fails
Application Insights is the documented unified observability experience for AI agents across Foundry, Copilot Studio, and third-party frameworks. Bypassing it forfeits the prebuilt agent dashboards. The documented path is to start tracing in Foundry, then add the Azure Monitor OpenTelemetry Distro with the Foundry SDK.
Design Identity, Governance, and Monitoring Solutions
Tempting but wrong
Setting an Azure SQL Database failover group to Microsoft-managed (automatic) failover lets business owners keep control over when failover happens.
Why it fails
Microsoft-managed failover only triggers during widespread regional outages and removes human control over the timing of failover. To keep the decision with the business, use customer-managed (manual) failover, which the documentation recommends so the business retains control.
Design Data Storage Solutions
Tempting but wrong
Switching an Azure VM to Hyper-V replication lets Azure Site Recovery create crash-consistent recovery points every 30 seconds for that workload.
Why it fails
The 30-second cadence belongs to the Hyper-V-to-Azure (H2A) replication path, not to Azure VM A2A. For A2A workloads the crash-consistent interval is fixed at five minutes and cannot be modified by changing replication type.
Design Business Continuity Solutions
Tempting but wrong
The 230-second HTTP response limit is the overall execution timeout for any Azure Functions trigger regardless of plan.
Why it fails
230 seconds is only the HTTP-trigger response limit imposed by the Azure Load Balancer; it is not the overall function timeout for non-HTTP triggers. On the legacy Consumption plan the documented execution maximum is 10 minutes with a 5-minute default.
Design Infrastructure Solutions
Tempting but wrong
The Microsoft Agent Framework is the way to trace a Foundry-hosted AI agent instead of using Application Insights.
Why it fails
The Microsoft Agent Framework is the documented self-hosting path. For Foundry-hosted agents the unified APM experience lives in Application Insights, reached by setting up tracing in Foundry and adding the Azure Monitor OpenTelemetry Distro with the Foundry SDK.
Design Identity, Governance, and Monitoring Solutions
Tempting but wrong
A zero-hour grace period on Microsoft-managed Azure SQL failover gives you human-controlled failover timing because nothing waits.
Why it fails
A zero-hour grace period still hands the failover decision to Microsoft; it only removes the waiting period, not the automation. Human-controlled timing requires customer-managed (manual) failover instead.
Design Data Storage Solutions
Tempting but wrong
Enabling multi-VM consistency on an ASR replication group overrides the default five-minute crash-consistent recovery-point interval.
Why it fails
Multi-VM consistency only synchronises recovery points across the machines in the group so they share a consistent point in time. It does not change the underlying crash-consistent interval, which stays fixed at five minutes.
Design Business Continuity Solutions
Key terms
Exam-day rules
- Extract every constraint before judging the options. Case studies and scenarios stack identity, governance, recovery, cost, and operational limits; the answer is the design that satisfies all of them, not just the most obvious one.
- When two designs both work, default to the managed, documented pattern at the least overhead. Microsoft writes the exam around its well-architected guidance, so reach for the heavier or less obvious option only when the scenario names a reason.
- Match the tier to the feature. Many traps hinge on a capability being tier-exclusive: multi-region API Management is Classic Premium only, Hyperscale is Azure SQL Database only, so check the tier supports the requirement before choosing it.
- Respect the documented platform limit. The exam loves hard numbers: 500 Recovery Services vaults per subscription, ASR crash-consistent points fixed at five minutes, one of each destination type per diagnostic setting. If an option ignores a stated limit, it is wrong.
- Map recovery objectives straight to the design. An RPO or RTO target picks the zone-versus-region choice and the replication mode: synchronous across zones for high-priority data, asynchronous across regions for the rest, customer-managed failover when humans must control timing.
Revision schedule
- Day 1Map the blueprint and book a date
- Week 1Build the design-decision maps
- Weeks 1 to 3Go deep on Infrastructure (Domain 4)
- Weeks 3 to 4Lock Identity, Governance, and Monitoring (Domain 1)
- Week 4Cover Data Storage and Business Continuity (Domains 2 and 3)