Microsoft AZ-900: Cost Management & Service Economics — Study Guide
Part of the Microsoft Azure AZ-900 — Study Guide. Practice with verified answers in the Microsoft exam hub, or take timed practice tests on ExamRoll.io.
Cloud economics in Azure centers on converting capital expenditure into operational expenditure and aligning spend to actual usage. Instead of overprovisioning for peak capacity, resources scale up and down to match demand, and you pay only for what you consume. This elasticity is the foundation for cost optimization: deploy the right resource, at the right size, in the right region, and only for as long as needed. Compute charges accrue while a virtual machine is running; when it is stopped (deallocated), compute charges cease but storage and networking costs remain. Data transfer out of Azure (egress) is billable, while most inbound traffic is not. Selecting cost-aware architectures—such as using managed services where appropriate, right-sizing compute, and planning data movement—yields predictable spending without sacrificing performance or resiliency.
Pricing models and commitment options
Azure offers multiple purchasing models to balance flexibility and savings. Pay-as-you-go is the default, on-demand pricing model with no commitment and maximal agility. It suits unpredictable or short-lived workloads, pilots, and development environments where shutting down resources quickly is common. Reserved Virtual Machine Instances provide significant discounts in exchange for a one- or three-year commitment on a specific VM family, region, and scope. They are ideal for steady-state workloads with predictable usage, such as line-of-business application servers or always-on databases. Reservations can often be exchanged within the same product family to accommodate evolving needs, and early cancellations are possible subject to policy and fees, making them less rigid than traditional capital purchases. Spot Virtual Machines deliver the lowest prices for interruptible workloads by using surplus Azure capacity. They are evicted when Azure needs capacity or when the market price exceeds your max price, so they are best for batch processing, CI/CD runners, rendering, and other fault-tolerant tasks. Production SLAs do not apply to Spot; designs should anticipate restarts and evictions. Azure Savings Plans for Compute offer discounts by committing to a fixed hourly spend over one or three years. Unlike reservations, savings plans automatically apply across eligible compute services and regions to maximize coverage, trading absolute peak discounts for broader flexibility. They reduce management overhead for diverse fleets spanning multiple VM families, container hosts, and function plans.
- Pay-as-you-go
- Commitment: None
- Typical Savings: —
- Flexibility: Highest flexibility
- Best for: Unpredictable/short-lived workloads
- Key Constraints: No discount; full on-demand rates
- Reserved VM Instances (1/3-year)
- Commitment: Term-based on VM family/region/scope
- Typical Savings: Up to ~72% vs PAYG (with Azure Hybrid Benefit)
- Flexibility: Exchange within product family; limited cancellation
- Best for: Steady-state VMs with consistent usage
- Key Constraints: Scoped to specific families/regions; manage coverage
- Spot VMs
- Commitment: None (capacity-dependent)
- Typical Savings: Up to ~90% vs PAYG
- Flexibility: Scale sets support eviction policies
- Best for: Batch, dev/test, stateless services
- Key Constraints: No SLA; evictions at any time
- Azure Savings Plan for Compute (1/3-year)
- Commitment: Hourly spend commitment
- Typical Savings: Typically between PAYG and RI peak savings
- Flexibility: Applies across services/regions to eligible compute
- Best for: Mixed fleets with variable usage patterns
- Key Constraints: Discount applies up to committed spend; overage at PAYG
Pricing drivers and estimation tools
Workload cost varies by region due to infrastructure economics and energy markets. Selecting a region near users reduces latency but may affect price; compliance or data residency may also dictate region choice. VM size and operating system impact cost: larger vCPU/RAM configurations cost more, and Windows licensing is embedded unless you use Azure Hybrid Benefit to bring eligible licenses. Storage pricing depends on performance tier and redundancy. Premium SSD and Ultra Disks deliver high IOPS/throughput at higher cost; redundancy options like LRS, ZRS, GRS, and GZRS incrementally add resiliency and, for geo-redundant options, additional storage and potential data egress during read access. Data egress to the internet or between regions is billable and should be minimized through caching, CDN, or locality-aware architectures. The Azure Pricing Calculator estimates monthly costs for proposed designs. Build a representative bill by selecting services, regions, sizes, and expected usage, then layer in options such as reservations, savings plans, Azure Hybrid Benefit, and dev/test pricing. Use it collaboratively with architecture, operations, and finance to iterate on design alternatives and understand cost sensitivity. The TCO Calculator compares the cost of running on-premises infrastructure to Azure over time. It models servers, storage, networking, facilities, power, cooling, and IT labor against Azure equivalents, revealing the operational expenditure profile and potential savings from elasticity and managed services. Calibrate inputs such as hardware refresh cycles, utilization, and growth rates to avoid underestimating on-premises carrying costs and to quantify benefits like retiring backup hardware or consolidating disaster recovery.
- Region
- Impact on Cost: Varies by market
- Considerations: Latency, data residency, price differences
- Mitigations/Levers: Place compute near data; compare nearby regions
- VM size/OS
- Impact on Cost: Larger = higher; Windows includes license
- Considerations: Azure Hybrid Benefit, Reserved Instances, Savings Plan
- Mitigations/Levers: Right-size; use B-series for bursty loads
- Storage tier/redundancy
- Impact on Cost: Premium/Ultra and geo-redundancy cost more
- Considerations: IOPS/throughput and durability targets
- Mitigations/Levers: Tiering, lifecycle management, fit-for-purpose disks
- Data egress
- Impact on Cost: Internet and cross-region egress billed
- Considerations: CDN, traffic patterns, DR read-access
- Mitigations/Levers: CDN/Front Door; keep data close; compress/de-dupe
Azure Cost Management + Billing
Azure Cost Management + Billing provides visibility, control, and accountability from subscription to enterprise agreement scopes. Cost analysis offers curated views by service, resource, tag, resource group, subscription, and management group, with filters and groupings to track trends and identify anomalies. Use saved views and scheduled email delivery to keep stakeholders informed. Budgets enforce spend guardrails with thresholds that trigger alerts or automated actions through Action Groups, Logic Apps, or Functions. Set monthly, quarterly, or annual budgets at the appropriate scope and align them to fiscal calendars. Pair budgets with Azure Policy to prevent resource creation outside approved regions or SKUs when costs approach limits. Exports deliver raw cost and usage data to a storage account on a schedule in CSV or Parquet formats. Downstream tools like Power BI, FinOps platforms, or data lake pipelines can consume these feeds for advanced analytics, showback/chargeback, and forecasting. For consolidated environments, scopes can align to management groups or billing accounts so shared services and enterprise discounts are reflected consistently. Cost allocation features allow distributing shared costs—such as hub networking, security tooling, or monitoring—across consuming teams using rules based on percentages or drivers like tag values. Implement consistent tagging and resource hierarchies so allocations are transparent, repeatable, and auditable.
- Cost analysis
- Scope: Sub, RG, MG, billing account
- Primary Use: Visibility and trend analysis
- Operational Pattern: Saved views; scheduled insights to owners
- Budgets + alerts
- Scope: Sub, RG, MG
- Primary Use: Spend guardrails and automation triggers
- Operational Pattern: Threshold alerts; policy tie-ins; action groups
- Exports
- Scope: Sub, billing account
- Primary Use: Data integration for BI/FinOps
- Operational Pattern: Daily scheduled CSV/Parquet to Storage
- Cost allocation
- Scope: Billing and management scopes
- Primary Use: Chargeback/showback for shared services
- Operational Pattern: Percent/driver-based allocation rules
Optimization techniques and governance controls
Azure Advisor surfaces tailored recommendations across cost, performance, reliability, and security. The cost category flags idle or underutilized resources, unattached disks, and low-usage public IPs. Align Advisor with operational reviews, and convert insights into backlog items with owners and due dates. Pair with Azure Monitor to validate utilization baselines before making changes. Right-sizing reduces waste by matching resources to observed demand. Resize VMs based on CPU, memory, and disk metrics; consider burstable B-series for spiky workloads and scale sets with autoscale for horizontal elasticity. For storage, select the lowest tier that satisfies IOPS/throughput and latency requirements, and enable lifecycle management to tier or delete cold data. Evaluate managed PaaS services such as Azure App Service or Azure SQL Database to offload undifferentiated operations and align capacity to consumption-based models. Use automated schedules to control non-production runtime. The Start/Stop VMs v2 solution in Azure Automation orchestrates time-based shutdown and startup across subscriptions and resource groups, respecting dependencies. VM auto-shutdown is also available per-VM for simple nightly shutdowns. Combine schedules with dev/test subscriptions and images to minimize runtime while preserving developer productivity. Commitment coverage should be intentional. Track reservation and savings plan utilization; adjust scope and exchanges to maintain high coverage without oversubscription. Apply Azure Hybrid Benefit where eligible to reduce Windows Server and SQL Server costs. For data, reduce egress by placing compute near data, using Azure CDN or Front Door, and avoiding unnecessary cross-region replication for chatty services.
- Advisor-driven cleanup
- Primary Lever: Remove idle/unattached resources
- Typical Savings: Low-hanging 10–25% on waste
- Notes: Operationalize into monthly cadence
- Right-sizing
- Primary Lever: Resize/scale-out/in
- Typical Savings: 20–50% on oversized compute
- Notes: Use Monitor baselines before changes
- Start/Stop schedules
- Primary Lever: Reduce non-prod runtime
- Typical Savings: 30–60% in dev/test
- Notes: Automation solution or per-VM auto-shutdown
- Commitment coverage
- Primary Lever: Reservations/Savings Plans
- Typical Savings: 20–60% vs PAYG
- Notes: Continuously tune utilization
- Hybrid licensing
- Primary Lever: Azure Hybrid Benefit
- Typical Savings: Up to ~40% on Windows/SQL
- Notes: Requires eligible licenses
Cost allocation structure: tags, resource groups, subscriptions, and management groups
A clear resource hierarchy enables granular accountability and effective policy enforcement. Design management groups to mirror organizational structure or cost centers at a high level, then place subscriptions under them to separate environments (production vs. non-production), business units, or regulatory boundaries. Subscriptions provide quotas, access boundaries, and billing separation suitable for chargeback. Resource groups organize related resources with a shared lifecycle—for example, an application’s compute, storage, and networking. They are the operational unit for deployment and cleanup, making them a practical lens for cost analysis when each application or environment has its own group. Enforce standards with Azure Policy to require tags such as costCenter, environment, application, and owner on all resources and resource groups. Tags are key-value pairs that travel with resources and appear in Cost Management. Standardize tag taxonomies so reporting remains consistent across teams and tools. Use deployment pipelines to stamp required tags automatically; deny creations lacking required tags. Cost allocation rules and exports can then reliably attribute shared or untagged spend to the right consumers. This layered structure—management groups, subscriptions, resource groups, and tags—enables both top-down governance and bottom-up accountability. It supports showback/chargeback models, enforces regional and SKU guardrails, and aligns budget ownership with the teams that can act on optimization opportunities.
- Management group
- Purpose: Org-wide hierarchy
- Cost Role: Roll-up, policy, budgets at scale
- Governance Mechanisms: Azure Policy, RBAC, budgets at MG scope
- Subscription
- Purpose: Isolation and billing boundary
- Cost Role: Chargeback/showback, quotas
- Governance Mechanisms: RBAC, policy assignments, budget scopes
- Resource group
- Purpose: Lifecycle grouping
- Cost Role: App/environment cost lens
- Governance Mechanisms: ARM deployments, cleanup, policy
- Tags
- Purpose: Metadata for attribution
- Cost Role: Fine-grained allocation and reporting
- Governance Mechanisms: Policy for required tags; pipeline stamping
Practical Problem: Contoso Retail: Bringing spend under control without slowing delivery
Scenario: Contoso Retail runs 120 VMs across multiple regions for e-commerce and analytics. Monthly Azure invoices have climbed unpredictably, with spikes at quarter end. Development teams maintain separate non-production environments that often run 24x7. Leadership needs predictable spend, better allocation by business unit, and concrete savings without disrupting the peak shopping season.
Challenge: Stabilize monthly costs and attribute spend to business units while maintaining performance and availability for critical retail workloads.
Recommended Approach:
- Establish governance hierarchy: create management groups for Corporate, E-commerce, and Analytics; move existing subscriptions under the appropriate management group; create separate non-production subscriptions for each unit.
- Enforce tagging: deploy Azure Policy at the management group level to require costCenter, environment, application, and owner tags on all resources and resource groups; remediate non-compliant resources via bulk tagging scripts.
- Baseline and analyze: use Cost Analysis with groupings by subscription, resource group, and tags to identify top spend drivers; enable daily cost exports to a storage account and connect Power BI for trend dashboards.
- Set budgets and alerts: create monthly budgets per subscription with 50/75/90% thresholds; wire alerts to Teams and a Logic App that opens backlog items for the owning team.
- Optimize runtime: deploy Start/Stop VMs v2 to schedule non-production VMs off 7pm–7am weekdays and all weekend; configure per-VM auto-shutdown where granular control is sufficient.
- Right-size and clean up: act on Azure Advisor recommendations to resize underutilized VMs, remove unattached disks, and downgrade overprovisioned managed disks; implement storage lifecycle management to tier infrequently accessed blobs.
- Apply commitment discounts: purchase a 1-year Azure Savings Plan for Compute sized to 70% of observed average hourly compute; add targeted 1-year Reserved VM Instances for two always-on database VM families with stable demand.
- Reduce egress and redundancy costs: colocate analytics compute with data in the same regions; review geo-redundant storage and switch non-critical datasets from GRS to ZRS or LRS where business RPO/RTO allows; front e-commerce static assets with Azure CDN.
- Monitor utilization and coverage: review reservation and savings plan utilization monthly; adjust scopes and exchanges to maintain high coverage without overcommitting.
- Iterate allocations: implement cost allocation rules to distribute shared hub networking and security services to consuming cost centers based on tagged VM counts or vCPU usage; refine quarterly.
Azure Rationale: This approach builds a governance foundation for accountability, then uses Cost Management to establish visibility and guardrails. Scheduling and right-sizing deliver immediate savings without architectural risk, while savings plans and selective reservations lock in discounts for predictable baseload. Tag-driven allocation and exports provide transparent chargeback, steering teams toward continuous optimization while preserving performance during peak periods.
← Identity · All domains · Governance →
Practice these questions → · Timed practice on ExamRoll.io →
Pass the whole exam — not just this question
You found this answer. Get every verified question and explanation in one place, and save hours of prep. Free to start.
Pass your exam →