Cost Optimization Pillar

TL;DR

The Cost Optimization pillar focuses on maximizing the value delivered by your cloud spend. Key concepts:

• Right-sizing: Match resources to actual workload needs
• Commitment discounts: Use reservations and savings plans for predictable workloads
• Spot instances: Leverage unused capacity for fault-tolerant workloads
• FinOps practices: Implement financial accountability and governance
• Continuous optimization: Regularly review and adjust spending

---

Design Principles

Core Cost Optimization Principles

Principle	Description	Implementation
Choose the right resources	Match SKU to requirements	Right-sizing analysis
Set up budgets and alerts	Proactive cost monitoring	Azure Cost Management
Dynamically allocate resources	Scale with demand	Auto-scaling, serverless
Optimize workloads	Improve efficiency	Code optimization, caching
Continuously monitor	Track and adjust	Cost reviews, anomaly detection

Cost Optimization Lifecycle

---

Azure Pricing Models

Pricing Model Comparison

Model	Discount	Commitment	Best For
Pay-as-you-go	0%	None	Variable workloads, testing
Reserved Instances	Up to 72%	1 or 3 years	Steady-state production
Savings Plans	Up to 65%	1 or 3 years	Flexible compute usage
Spot VMs	Up to 90%	None (can be evicted)	Fault-tolerant batch jobs
Dev/Test Pricing	Up to 55%	VS subscription	Development environments
Hybrid Benefit	Up to 85%	Existing licenses	Windows/SQL migrations

When to Use Each Model

---

Right-Sizing Strategies

Identifying Right-Sizing Opportunities

Right-Sizing Guidelines

Metric	Threshold	Action
CPU < 20% average	Underutilized	Downsize or consolidate
CPU > 80% average	Constrained	Upsize or scale out
Memory < 30% average	Over-provisioned	Choose smaller SKU
Memory > 90% average	Memory pressure	Add memory or optimize
Disk IOPS < 50%	Over-provisioned	Use standard storage

Azure Advisor Recommendations

# Get cost recommendations from Azure Advisor
az advisor recommendation list --category Cost --output table

# Get specific VM right-sizing recommendations
az advisor recommendation list \
    --category Cost \
    --query "[?shortDescription.problem=='Right-size or shutdown underutilized virtual machines']"

---

Reserved Instances and Savings Plans

Reserved Instances

RI vs Savings Plans

Feature	Reserved Instances	Savings Plans
Flexibility	Specific SKU and region	Any SKU, any region
Discount	Up to 72%	Up to 65%
Scope	Subscription or shared	Subscription or shared
Exchange	Yes, with restrictions	No
Best for	Known, stable workloads	Dynamic compute needs

Calculating RI Savings

Monthly Pay-as-you-go cost: $1,000
3-year RI discount: 60%
Monthly RI cost: $400

Monthly savings: $600
Annual savings: $7,200
3-year savings: $21,600
Break-even: ~5 months

---

Spot VMs

Spot VM Use Cases

Use Case	Suitability	Reason
Batch processing	Excellent	Can restart interrupted jobs
CI/CD agents	Good	Short-lived, replaceable
Dev/Test	Good	Non-critical workloads
Rendering	Excellent	Parallelizable, fault-tolerant
Machine Learning	Good	Checkpoint-based training
Production web	Poor	Eviction causes downtime

Spot VM Configuration

// Bicep - Create Spot VM
resource spotVM 'Microsoft.Compute/virtualMachines@2023-07-01' = {
  name: 'spot-vm'
  location: location
  properties: {
    hardwareProfile: {
      vmSize: 'Standard_D4s_v3'
    }
    priority: 'Spot'
    evictionPolicy: 'Deallocate' // or 'Delete'
    billingProfile: {
      maxPrice: 0.05 // Max price per hour, -1 for market price
    }
    // ... other configuration
  }
}

Handling Spot Evictions

// C# - Check for Spot VM eviction using Scheduled Events
public async Task<bool> CheckForEviction()
{
    var client = new HttpClient();
    var response = await client.GetAsync(
        "http://169.254.169.254/metadata/scheduledevents?api-version=2020-07-01",
        new Dictionary<string, string> { { "Metadata", "true" } });
    
    var events = JsonSerializer.Deserialize<ScheduledEvents>(
        await response.Content.ReadAsStringAsync());
    
    return events.Events.Any(e => e.EventType == "Preempt");
}

---

Azure Cost Management

Cost Management Features

Setting Up Budgets

# Create a budget with Azure CLI
az consumption budget create \
    --budget-name "Monthly-Budget" \
    --amount 10000 \
    --category Cost \
    --time-grain Monthly \
    --start-date 2024-01-01 \
    --end-date 2024-12-31 \
    --resource-group myRG

# Create budget alert
az consumption budget create \
    --budget-name "Alert-Budget" \
    --amount 5000 \
    --category Cost \
    --time-grain Monthly \
    --notifications '{
        "Actual_GreaterThan_80_Percent": {
            "enabled": true,
            "operator": "GreaterThan",
            "threshold": 80,
            "contactEmails": ["admin@example.com"],
            "thresholdType": "Actual"
        }
    }'

Cost Analysis Queries

// KQL - Analyze costs by resource type
AzureCostManagement
| where TimeGenerated > ago(30d)
| summarize TotalCost = sum(Cost) by ResourceType
| order by TotalCost desc
| take 10

// Costs by tag
AzureCostManagement
| where TimeGenerated > ago(30d)
| extend CostCenter = tostring(Tags["CostCenter"])
| summarize TotalCost = sum(Cost) by CostCenter
| order by TotalCost desc

---

FinOps Practices

FinOps Framework

FinOps Maturity Model

Stage	Characteristics	Focus
Crawl	Basic visibility, reactive	Understand spending
Walk	Proactive optimization, some automation	Reduce waste
Run	Full automation, predictive	Maximize value

Cost Allocation Strategy

Dimension	Purpose	Example
Environment	Separate prod/dev costs	env:production
Cost Center	Business unit allocation	costcenter:engineering
Project	Project-level tracking	project:customer-portal
Owner	Accountability	owner:team-alpha
Application	Application costs	app:order-service

---

Tagging Strategy

Required Tags

{
  "Environment": "production",
  "CostCenter": "CC-12345",
  "Owner": "platform-team@company.com",
  "Project": "customer-portal",
  "Application": "order-service",
  "CreatedBy": "terraform",
  "CreatedDate": "2024-01-15"
}

Enforcing Tags with Azure Policy

// Azure Policy - Require CostCenter tag
{
  "mode": "Indexed",
  "policyRule": {
    "if": {
      "allOf": [
        {
          "field": "tags['CostCenter']",
          "exists": false
        },
        {
          "field": "type",
          "notEquals": "Microsoft.Resources/subscriptions/resourceGroups"
        }
      ]
    },
    "then": {
      "effect": "deny"
    }
  }
}

# Apply tag to all resources in resource group
az tag create --resource-id /subscriptions/{sub}/resourceGroups/{rg} \
    --tags CostCenter=CC-12345 Environment=production

# List untagged resources
az resource list --query "[?tags.CostCenter==null].{Name:name, Type:type}" -o table

---

Storage Cost Optimization

Storage Tier Comparison

Tier	Access Frequency	Cost (per GB)	Access Cost
Hot	Frequent	$$$	$
Cool	Infrequent (30+ days)	$$	$$
Cold	Rare (90+ days)	$	$$$
Archive	Rarely (180+ days)	¢	$$$$

Lifecycle Management

// Storage Lifecycle Policy
{
  "rules": [
    {
      "name": "MoveToCool",
      "type": "Lifecycle",
      "definition": {
        "filters": {
          "blobTypes": ["blockBlob"],
          "prefixMatch": ["logs/"]
        },
        "actions": {
          "baseBlob": {
            "tierToCool": { "daysAfterModificationGreaterThan": 30 },
            "tierToArchive": { "daysAfterModificationGreaterThan": 90 },
            "delete": { "daysAfterModificationGreaterThan": 365 }
          }
        }
      }
    }
  ]
}

---

Compute Cost Optimization

Auto-Scaling Configuration

// Bicep - Autoscale settings for App Service
resource autoscale 'Microsoft.Insights/autoscalesettings@2022-10-01' = {
  name: 'autoscale-webapp'
  location: location
  properties: {
    targetResourceUri: appServicePlan.id
    enabled: true
    profiles: [
      {
        name: 'Default'
        capacity: {
          minimum: '1'
          maximum: '10'
          default: '2'
        }
        rules: [
          {
            metricTrigger: {
              metricName: 'CpuPercentage'
              metricResourceUri: appServicePlan.id
              timeGrain: 'PT1M'
              statistic: 'Average'
              timeWindow: 'PT5M'
              timeAggregation: 'Average'
              operator: 'GreaterThan'
              threshold: 70
            }
            scaleAction: {
              direction: 'Increase'
              type: 'ChangeCount'
              value: '1'
              cooldown: 'PT5M'
            }
          }
          {
            metricTrigger: {
              metricName: 'CpuPercentage'
              metricResourceUri: appServicePlan.id
              timeGrain: 'PT1M'
              statistic: 'Average'
              timeWindow: 'PT5M'
              timeAggregation: 'Average'
              operator: 'LessThan'
              threshold: 30
            }
            scaleAction: {
              direction: 'Decrease'
              type: 'ChangeCount'
              value: '1'
              cooldown: 'PT5M'
            }
          }
        ]
      }
    ]
  }
}

Serverless Cost Benefits

Scenario	Traditional	Serverless	Savings
Low traffic API	24/7 VM ($100/mo)	Pay per request ($5/mo)	95%
Batch processing	Dedicated VM ($200/mo)	Functions ($20/mo)	90%
Event processing	Always-on service	Event-driven ($10/mo)	85%

---

Cost Optimization Checklist

Planning Phase

• Estimate costs using Azure Pricing Calculator
• Set budgets and alerts
• Define tagging strategy
• Plan for reserved capacity

Implementation Phase

• Apply required tags to all resources
• Configure auto-scaling
• Use appropriate storage tiers
• Implement lifecycle policies

Operations Phase

• Review Azure Advisor recommendations weekly
• Analyze cost trends monthly
• Right-size underutilized resources
• Purchase/renew reservations as needed

Governance Phase

• Enforce tagging with Azure Policy
• Implement cost allocation
• Conduct regular cost reviews
• Train teams on cost awareness

---

Assessment Questions

Area	Question
Visibility	Can you see costs by team/project/application?
Budgets	Do you have budgets and alerts configured?
Right-sizing	When did you last review resource utilization?
Reservations	Are predictable workloads covered by RIs?
Tagging	Are all resources properly tagged?
Automation	Do you auto-scale based on demand?
Storage	Are you using appropriate storage tiers?
Governance	Who is accountable for cloud costs?

---

Key Takeaways

1. Right-size first: Don't pay for resources you don't use
2. Commit to save: Use reservations for predictable workloads
3. Tag everything: Enable cost allocation and accountability
4. Automate scaling: Match capacity to demand
5. Review regularly: Cost optimization is continuous

---

Resources

• Azure Cost Management Documentation
• Azure Pricing Calculator
• Azure Advisor Cost Recommendations
• FinOps Foundation
• Azure Reserved Instances
• Azure Hybrid Benefit