Cost-Aware Architecture Design Patterns

Cost-aware architecture design is the practice of building cloud solutions that optimize spending without sacrificing performance, security, or reliability. In Azure, this means selecting the right services, sizes, and configurations to match workload demands while using tools like Azure Cost Management + Billing, Azure Advisor, and Azure Policy to monitor and control costs. The AZ-305 exam expects you to recommend cost-optimized designs using specific patterns such as right-sizing, reserved instances, spot VMs, and auto-scaling.

Microsoft defines five pillars of the Azure Well-Architected Framework: Reliability, Security, Cost Optimization, Operational Excellence, and Performance Efficiency. Cost Optimization is the pillar most directly tested in objective 1.3. The exam will present scenarios where you must choose between multiple cost-saving options, such as reserved vs. pay-as-you-go, or standard vs. low-priority VMs. You must also understand how to implement governance policies that prevent cost overruns, such as Azure Policy for allowed SKUs and tagging requirements.

Azure Cost Management + Billing: This is the primary tool for monitoring, analyzing, and optimizing Azure costs. It provides: - Cost Analysis: View and filter costs by subscription, resource group, resource type, or tags. You can create custom views and export data to Excel or Power BI. - Budgets: Set spending limits and receive alerts when costs exceed thresholds (e.g., 50%, 80%, 100% of budget). Budgets can trigger automation, such as disabling a resource or sending an email. - Recommendations: Azure Advisor provides cost recommendations, such as right-sizing underutilized VMs or purchasing reserved instances. These recommendations are based on 30 days of usage data. - Cost Alerts: Three types: budget alerts (when spending reaches a threshold), credit alerts (when Azure credits are consumed), and department spending quota alerts (for EA customers).

Azure Policy for Cost Governance: Azure Policy can enforce cost-related rules, such as:

Azure Reservations and Savings Plans: These offer significant discounts (up to 72%) compared to pay-as-you-go pricing for one-year or three-year commitments. Reservations apply to specific VM sizes, SQL Database, Cosmos DB, and other services. Savings Plans are more flexible, covering any compute service across regions. The exam tests when to recommend reservations vs. pay-as-you-go based on steady-state usage.

Azure Spot VMs: These are unused compute capacity offered at up to 90% discount, but they can be evicted at any time (with 30-second notice) when Azure needs the capacity back. Ideal for batch processing, dev/test, or stateless workloads that can tolerate interruptions. The exam expects you to know the eviction policy and use cases.

Auto-Scaling: Scale out (add instances) and scale in (remove instances) based on metrics like CPU, memory, or queue length. This ensures you only pay for what you use. The exam tests how to configure autoscale settings and when to use predictive autoscale (based on historical patterns).

Right-Sizing: This involves adjusting VM sizes, storage tiers, and other resources to match actual usage. Azure Advisor analyzes CPU, memory, and disk I/O over 30 days and recommends downsizing or shutting down idle VMs. For example, a VM running at 5% CPU for a month should be downsized from Standard_D4s_v3 to Standard_D2s_v3, saving 50% cost. The exam will ask you to interpret Advisor recommendations.

Reserved Instances (RIs): You commit to a specific VM size in a region for one or three years. The discount applies automatically to matching VMs. There are three payment options: upfront (highest discount), partial upfront, and monthly (lowest discount). The exam may present a scenario where a customer has predictable, steady-state workloads and ask which RI payment option is most cost-effective.

Hybrid Benefit: If you have Windows Server or SQL Server licenses with Software Assurance, you can use them in Azure to reduce costs. This applies to VMs and Azure SQL Database. The exam tests when to enable Hybrid Benefit (e.g., for Windows VMs to save on licensing) and the prerequisites (Software Assurance or subscription licenses).

Azure Budgets and Action Groups: You can create budgets at management group, subscription, or resource group scope. When a budget threshold is exceeded, an action group can send email, SMS, or trigger an automation runbook or webhook. For example, a budget at 80% could send an email to the finance team; at 100%, it could trigger a runbook that shuts down non-critical VMs.

Tagging Strategy: Tags are metadata key-value pairs applied to resources. A consistent tagging strategy enables cost allocation and chargeback. Common tags: CostCenter, Department, Environment (Prod/Dev/Test), Owner, Project. Azure Policy can enforce tagging and inherit tags from resource groups. The exam will ask you to design a tagging strategy for cost reporting.

Using Azure CLI to view costs:

az consumption budget list --subscription <subscription-id>
az consumption usage list --subscription <subscription-id> --start-date 2024-01-01 --end-date 2024-01-31

Using Azure Policy to enforce allowed VM SKUs:

{
  "policyRule": {
    "if": {
      "allOf": [
        {
          "field": "type",
          "equals": "Microsoft.Compute/virtualMachines"
        },
        {
          "not": {
            "field": "Microsoft.Compute/virtualMachines/sku.name",
            "in": ["Standard_B2s", "Standard_D2s_v3"]
          }
        }
      ]
    },
    "then": {
      "effect": "deny"
    }
  }
}

Cost-aware design interacts with: - Azure Advisor: Provides cost recommendations that you can implement via policies or manual actions. - Azure Monitor: Metrics and logs feed into autoscale decisions and cost analysis. - Azure Automation: Runbooks can automate cost-saving actions like shutting down VMs during off-hours. - Azure Blueprints: Can include policy assignments and RBAC roles to enforce cost governance at scale.

A large e-commerce company runs hundreds of VMs across multiple regions for its web tier, application tier, and databases. They were spending $500,000/month on Azure. The finance team needed to allocate costs to business units. They implemented a comprehensive tagging strategy with tags like 'CostCenter', 'Environment', and 'Application'. Azure Policy enforced tagging on all new resources. They used Azure Cost Management to create budgets per CostCenter and set alerts at 80% and 100%. They also purchased reserved instances for their steady-state database VMs, saving 40%. For the web tier, they implemented autoscale based on CPU and used spot VMs for batch processing jobs. Within six months, they reduced costs by 30% while maintaining performance.

A SaaS provider had unpredictable traffic patterns, making reservations risky. They used Azure Savings Plans (1-year term) to get a 20% discount on compute without committing to specific VM sizes. They also used Azure Policy to restrict VM SKUs to the B-series for dev/test and Dv3-series for production. They set up budgets at the subscription level with alerts that triggered an Azure Automation runbook to scale down non-critical resources when spending exceeded 90% of budget. They also enabled Azure Advisor cost recommendations and scheduled a monthly review to right-size VMs. This approach helped them stay within budget while handling traffic spikes.