Pricing Misconception — Azure Fundamentals (AZ-900)
The answer assumed a pricing model that doesn't apply. Reserved vs. on-demand, data transfer costs, or tier thresholds tripped you up.
Free Tier Doesn't Mean Free at Production Scale
The scenario mentions cost reduction and a candidate spots an answer referencing the Azure free tier or a zero-cost service. The instinct is right in direction but wrong in scope. Free tiers cap at usage thresholds; reserved instances require upfront commitment; support plans carry tiered SLA costs. The exam is testing whether you apply the actual billing constraint — not whether you pattern-match on the word 'free.'
The Scenario
A company wants to optimize costs for VMs running production workloads with predictable demand. You recommend Azure Reserved VM Instances for 3 years to maximize the discount. But the scenario says the company is "mid-migration and plans to rightsize workloads within the next quarter." Locking into a 3-year reservation for a D4s_v3 that will become a B2s after rightsizing means paying for 4x the compute you need. The correct answer is Azure Hybrid Benefit (if they have Windows Server licenses) combined with Azure Advisor rightsizing recommendations first, then reservations after workloads stabilize.
How to Spot It
- •Azure reservations lock you into specific VM families, regions, and terms. If the scenario mentions migration, modernization, or rightsizing, reservations are premature. The exam tests whether you stabilize workloads before committing.
- •Azure Functions Consumption plan charges per execution ($0.20 per million) plus execution time. For a workload that runs constantly at 100 executions per second, the Consumption plan costs more than a Premium plan with reserved instances. The exam tests whether you calculate consumption costs for high-frequency workloads.
- •Azure Storage lifecycle management can move data from Hot to Cool (50% cheaper) to Archive (90% cheaper). But rehydrating from Archive takes up to 15 hours with Standard priority. Questions about "rarely accessed data that must be available within minutes" test whether you know Archive retrieval times eliminate it — Cool tier is the answer.
Decision Rules
Whether consumption-based serverless pricing is the cost-optimal model for a steady-state, always-on workload, or whether a managed PaaS with predictable dedicated-tier pricing satisfies both the operational-overhead-minimization and cost-predictability constraints simultaneously.
Choose availability zones (intra-region, no cross-region data-transfer charges) over region pairs (inter-region geo-replication, which adds replication storage and egress costs) when the RTO target is satisfied by datacenter-level fault isolation and data must remain cost-efficiently within a single region.
Availability zone configuration and billing commitment tier are fully independent controls — zone placement must be explicitly declared at VM provisioning time via Azure Resource Manager regardless of whether the VM uses on-demand or reserved pricing, and selecting a Reserved Instance discount does not confer any fault-isolation or zone-distribution behavior.
Whether single-datacenter fault tolerance under a region-locked Azure Policy assignment is best achieved with availability zones (intra-region isolation, policy-compliant, no secondary-region cost) or region-pair replication (inter-region, expands residency scope beyond the policy-allowed region and incurs secondary compute and egress charges).
Whether Azure Policy's location-restriction scope inherently confers zone-redundancy within that region, or whether datacenter fault isolation requires an independent, explicit zone-placement declaration in the ARM deployment configuration.
Determine whether Azure Tags exert any fault-isolation or placement behavior (they do not) and whether zone-redundant deployment must be explicitly declared in ARM templates, which incurs additional per-zone VM compute charges the tag-based claim incorrectly dismisses.
Domain Coverage
Difficulty Breakdown
Related Patterns