Your Azure bill increased 50% last month. Identify waste and implement cost controls.

Step 1: Analyze Current Spending

# Get cost breakdown by resource group
az consumption usage list \
  --start-date 2024-01-01 \
  --end-date 2024-01-31 \
  --query "[].{ResourceGroup:resourceGroup,Cost:pretaxCost}" \
  --output table

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

# Export cost data for analysis
az costmanagement query \
  --type Usage \
  --scope "/subscriptions/{subscription-id}" \
  --timeframe MonthToDate \
  --dataset-grouping name=ResourceGroup type=Dimension \
  --dataset-aggregation '{"totalCost":{"name":"Cost","function":"Sum"}}'

Step 2: Implement Reserved Instances

// Reserved Instance savings calculator
// Standard D4s_v5 (4 vCPU, 16GB) in East US:
// - Pay-as-you-go: $140.16/month
// - 1-year reserved: $89.79/month (36% savings)
// - 3-year reserved: $57.67/month (59% savings)

// For 10 production VMs running 24/7:
// - Current: 10 × $140.16 = $1,401.60/month
// - With 3-year RI: 10 × $57.67 = $576.70/month
// - Annual savings: $9,899/year

// Purchase recommendations based on usage patterns
resource reservationOrder 'Microsoft.Capacity/reservationOrders@2022-11-01' = {
  name: 'ro-production-vms'
  location: 'global'
  properties: {
    reservedResourceType: 'VirtualMachines'
    billingScopeId: subscription().id
    term: 'P3Y'  // 3-year term
    billingPlan: 'Monthly'
    quantity: 10
    displayName: 'Production VM Reservations'
    appliedScopes: [subscription().id]
    appliedScopeType: 'Shared'  // Apply across subscriptions
    renew: true
  }
}

// Azure SQL Database reservations
// vCore reservations apply across all SQL products
resource sqlReservation 'Microsoft.Capacity/reservationOrders@2022-11-01' = {
  name: 'ro-sql-vcores'
  location: 'global'
  properties: {
    reservedResourceType: 'SqlDatabases'
    term: 'P1Y'
    quantity: 24  // Total vCores
    displayName: 'SQL vCore Reservations'
    appliedScopeType: 'Shared'
  }
}

Step 3: Right-Size Virtual Machines

// Enable Azure Monitor for VM metrics analysis
resource vmInsights 'Microsoft.Insights/dataCollectionRules@2022-06-01' = {
  name: 'dcr-vm-performance'
  location: location
  properties: {
    dataSources: {
      performanceCounters: [
        {
          name: 'VMPerformance'
          streams: ['Microsoft-Perf']
          samplingFrequencyInSeconds: 60
          counterSpecifiers: [
            '\\Processor Information(_Total)\\% Processor Time'
            '\\Memory\\% Committed Bytes In Use'
            '\\LogicalDisk(_Total)\\% Disk Read Time'
            '\\LogicalDisk(_Total)\\% Disk Write Time'
          ]
        }
      ]
    }
    destinations: {
      logAnalytics: [
        {
          workspaceResourceId: logAnalytics.id
          name: 'vmLogs'
        }
      ]
    }
  }
}

// KQL query to find oversized VMs
Perf
| where ObjectName == "Processor" and CounterName == "% Processor Time"
| where TimeGenerated > ago(30d)
| summarize AvgCPU = avg(CounterValue),
            MaxCPU = max(CounterValue),
            P95CPU = percentile(CounterValue, 95)
by Computer
| where P95CPU < 20  // VMs with P95 CPU < 20% are oversized
| order by AvgCPU asc

// Memory utilization
Perf
| where ObjectName == "Memory" and CounterName == "% Committed Bytes In Use"
| where TimeGenerated > ago(30d)
| summarize AvgMemory = avg(CounterValue),
            MaxMemory = max(CounterValue),
            P95Memory = percentile(CounterValue, 95)
by Computer
| where P95Memory < 40  // VMs with P95 Memory < 40% can be downsized

# Resize VM based on analysis
az vm resize \
  --resource-group rg-production \
  --name vm-web-01 \
  --size Standard_D2s_v5  # Downsize from D4s_v5

# Savings: D4s_v5 ($140/mo) → D2s_v5 ($70/mo) = 50% per VM

Step 4: Auto-Shutdown for Non-Production

// Auto-shutdown for dev/test VMs
resource autoShutdown 'Microsoft.DevTestLab/schedules@2018-09-15' = {
  name: 'shutdown-computevm-${vmName}'
  location: location
  properties: {
    status: 'Enabled'
    taskType: 'ComputeVmShutdownTask'
    dailyRecurrence: {
      time: '1900'  // 7 PM
    }
    timeZoneId: 'Eastern Standard Time'
    notificationSettings: {
      status: 'Enabled'
      timeInMinutes: 30
      emailRecipient: 'team@contoso.com'
    }
    targetResourceId: vm.id
  }
}

// Start VMs on schedule using Automation
resource automationRunbook 'Microsoft.Automation/automationAccounts/runbooks@2022-08-08' = {
  parent: automationAccount
  name: 'Start-DevVMs'
  location: location
  properties: {
    runbookType: 'PowerShell'
    logProgress: true
    logVerbose: false
    publishContentLink: {
      uri: 'https://raw.githubusercontent.com/contoso/runbooks/main/Start-DevVMs.ps1'
    }
  }
}

resource startSchedule 'Microsoft.Automation/automationAccounts/schedules@2022-08-08' = {
  parent: automationAccount
  name: 'StartDevVMsWeekday'
  properties: {
    startTime: '2024-01-01T08:00:00+00:00'
    frequency: 'Week'
    interval: 1
    timeZone: 'Eastern Standard Time'
    advancedSchedule: {
      weekDays: ['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday']
    }
  }
}

// Savings: Dev VMs running 10hrs/day × 5 days = 50hrs vs 720hrs
// = 93% cost reduction for dev VMs

Step 5: Optimize Azure SQL

-- Identify unused indexes
SELECT
    OBJECT_NAME(i.object_id) AS TableName,
    i.name AS IndexName,
    ius.user_seeks,
    ius.user_scans,
    ius.user_lookups,
    ius.user_updates
FROM sys.indexes i
JOIN sys.dm_db_index_usage_stats ius
    ON i.object_id = ius.object_id AND i.index_id = ius.index_id
WHERE OBJECTPROPERTY(i.object_id, 'IsUserTable') = 1
    AND ius.user_seeks = 0
    AND ius.user_scans = 0
    AND ius.user_lookups = 0
ORDER BY ius.user_updates DESC;

-- Check DTU/vCore utilization
SELECT
    AVG(avg_cpu_percent) as AvgCPU,
    MAX(avg_cpu_percent) as MaxCPU,
    AVG(avg_data_io_percent) as AvgIO,
    AVG(avg_memory_usage_percent) as AvgMemory
FROM sys.dm_db_resource_stats
WHERE end_time > DATEADD(day, -14, GETUTCDATE());

// Right-size based on analysis
resource sqlDatabase 'Microsoft.Sql/servers/databases@2023-05-01-preview' = {
  parent: sqlServer
  name: 'appdb'
  location: location
  sku: {
    // Before: Premium P4 (500 DTU) - $1,860/month
    // After: Standard S3 (100 DTU) - $150/month
    // Or: General Purpose 2 vCore - $370/month

    name: 'GP_S_Gen5'  // Serverless for variable workloads
    tier: 'GeneralPurpose'
    family: 'Gen5'
    capacity: 2
  }
  properties: {
    autoPauseDelay: 60  // Pause after 1 hour of inactivity
    minCapacity: 0.5     // Minimum 0.5 vCores when active
    zoneRedundant: false // Disable for non-prod
  }
}

// Use Elastic Pools for multiple databases
resource elasticPool 'Microsoft.Sql/servers/elasticPools@2023-05-01-preview' = {
  parent: sqlServer
  name: 'pool-shared'
  location: location
  sku: {
    name: 'GP_Gen5'
    tier: 'GeneralPurpose'
    family: 'Gen5'
    capacity: 4  // 4 vCores shared across databases
  }
  properties: {
    perDatabaseSettings: {
      minCapacity: 0
      maxCapacity: 2
    }
  }
}
// 10 databases × $370/month = $3,700
// vs Elastic Pool 4 vCore: $740/month = 80% savings

Step 6: Optimize AKS

// Right-size AKS node pools
resource aksCluster 'Microsoft.ContainerService/managedClusters@2023-05-01' = {
  name: aksName
  location: location
  properties: {
    agentPoolProfiles: [
      {
        name: 'system'
        count: 2  // Reduced from 3
        vmSize: 'Standard_D2s_v5'  // Reduced from D4s_v5
        mode: 'System'
        enableAutoScaling: true
        minCount: 2
        maxCount: 3
      }
      {
        name: 'workload'
        count: 3
        vmSize: 'Standard_D4s_v5'
        mode: 'User'
        enableAutoScaling: true
        minCount: 2
        maxCount: 10  // Scale up only when needed

        // Use spot instances for non-critical workloads
        scaleSetPriority: 'Spot'
        spotMaxPrice: -1  // Pay up to on-demand price
        scaleSetEvictionPolicy: 'Delete'

        nodeLabels: {
          'workload-type': 'batch'
        }
        nodeTaints: [
          'kubernetes.azure.com/scalesetpriority=spot:NoSchedule'
        ]
      }
    ]
  }
}

// Spot instance savings: ~60-90% vs regular VMs

# Kubernetes resource optimization
apiVersion: v1
kind: LimitRange
metadata:
  name: resource-limits
  namespace: production
spec:
  limits:
    - default:
        cpu: "500m"
        memory: "512Mi"
      defaultRequest:
        cpu: "100m"
        memory: "128Mi"
      type: Container

---
# Horizontal Pod Autoscaler
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: api-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: api
  minReplicas: 2
  maxReplicas: 20
  metrics:
    - type: Resource
      resource:
        name: cpu
        target:
          type: Utilization
          averageUtilization: 70

Step 7: Set Up Cost Alerts and Budgets

// Budget with alerts
resource budget 'Microsoft.Consumption/budgets@2023-05-01' = {
  name: 'monthly-budget'
  properties: {
    category: 'Cost'
    amount: 120000  // Target: $120K (20% reduction)
    timeGrain: 'Monthly'
    timePeriod: {
      startDate: '2024-01-01'
      endDate: '2025-12-31'
    }
    filter: {
      dimensions: {
        name: 'ResourceGroup'
        operator: 'In'
        values: ['rg-production', 'rg-staging', 'rg-development']
      }
    }
    notifications: {
      Actual_GreaterThan_80_Percent: {
        enabled: true
        operator: 'GreaterThan'
        threshold: 80
        contactEmails: ['finance@contoso.com', 'platform@contoso.com']
        contactRoles: ['Owner', 'Contributor']
        thresholdType: 'Actual'
      }
      Forecasted_GreaterThan_100_Percent: {
        enabled: true
        operator: 'GreaterThan'
        threshold: 100
        contactEmails: ['finance@contoso.com', 'cto@contoso.com']
        thresholdType: 'Forecasted'
      }
    }
  }
}

// Resource group level budget
resource rgBudget 'Microsoft.Consumption/budgets@2023-05-01' = {
  name: 'dev-budget'
  scope: resourceGroup('rg-development')
  properties: {
    category: 'Cost'
    amount: 5000
    timeGrain: 'Monthly'
    notifications: {
      Actual_GreaterThan_90_Percent: {
        enabled: true
        operator: 'GreaterThan'
        threshold: 90
        contactEmails: ['dev-lead@contoso.com']
      }
    }
  }
}

Step 8: Implement Cost Tagging

// Enforce tagging policy
resource taggingPolicy 'Microsoft.Authorization/policyDefinitions@2021-06-01' = {
  name: 'require-cost-tags'
  properties: {
    policyType: 'Custom'
    mode: 'Indexed'
    displayName: 'Require cost center and environment tags'
    policyRule: {
      if: {
        anyOf: [
          {
            field: 'tags[CostCenter]'
            exists: 'false'
          }
          {
            field: 'tags[Environment]'
            exists: 'false'
          }
          {
            field: 'tags[Owner]'
            exists: 'false'
          }
        ]
      }
      then: {
        effect: 'deny'
      }
    }
  }
}

// Apply tags to all resources
resource tagPolicy 'Microsoft.Resources/tags@2021-04-01' = {
  name: 'default'
  properties: {
    tags: {
      Environment: environment
      CostCenter: costCenter
      Owner: ownerEmail
      Project: projectName
      CreatedBy: 'Bicep'
      CreatedDate: utcNow('yyyy-MM-dd')
    }
  }
}

Cost Optimization Summary

Optimization Results:

BEFORE ($150,000/month):
├── VMs: $45,000
├── SQL: $30,000
├── Storage: $22,500
├── AKS: $25,500
├── App Services: $15,000
└── Other: $12,000

AFTER ($100,000/month):
├── VMs: $25,000 (-44%)
│   ├── Reserved instances: -$10,000
│   ├── Right-sizing: -$5,000
│   └── Auto-shutdown dev: -$5,000
├── SQL: $18,000 (-40%)
│   ├── Elastic pools: -$8,000
│   └── Serverless: -$4,000
├── Storage: $15,000 (-33%)
│   └── Lifecycle policies: -$7,500
├── AKS: $18,000 (-29%)
│   ├── Spot instances: -$5,000
│   └── Autoscaling: -$2,500
├── App Services: $12,000 (-20%)
│   └── Right-size non-prod: -$3,000
└── Other: $12,000

TOTAL SAVINGS: $50,000/month (33% reduction)
ANNUAL SAVINGS: $600,000