Skip to main content

09 - Cost Optimization

Tier selection, scaling strategies, and FinOps practices

💰 Tier Pricing Comparison​

TierBase Cost/MonthScale UnitsBest For
Developer~$50No SLADev/Test
Basic~$150FixedLow traffic
Standard~$700IncludedMedium workloads
Premium~$2,800Add-onEnterprise
Basic v2~$170Add-onModerate + VNet
Standard v2~$280Add-onHigh + VNet

💡 Premium costs ~4x Standard but includes zone redundancy, multi-region, and VNet

📊 Cost Decision Tree​

🎯 Right-Sizing Guidance​

Traffic-Based Selection​

Requests/SecondRecommended TierEstimated Cost
< 100Basic~$150/mo
100-500Standard~$700/mo
500-1,000Standard + Units~$1,000/mo
1,000-5,000Standard v2~$500/mo
> 5,000Premium~$3,500+/mo

Feature-Based Selection​

RequirementMinimum Tier
Development/TestingDeveloper
Basic productionBasic
Caching, autoscaleStandard
VNet integrationBasic v2 / Premium
Multi-regionPremium
Zone redundancyPremium
Self-hosted gatewayPremium

💡 Cost Saving Strategies​

1. Use Dev/Test Environments Wisely​

// Use Developer tier for non-production
param environment string = 'dev'

var skuName = environment == 'prod' ? 'Premium' : 'Developer'
var skuCapacity = environment == 'prod' ? 2 : 1

resource apim 'Microsoft.ApiManagement/service@2023-05-01-preview' = {
  name: 'apim-${environment}'
  location: location
  sku: {
    name: skuName
    capacity: skuCapacity
  }
}

2. Schedule Scale-Down (Non-Production)​

// Scale down after hours using automation
resource scaleDownSchedule 'Microsoft.Automation/automationAccounts/schedules@2023-11-01' = {
  name: 'apim-scale-down'
  parent: automationAccount
  properties: {
    frequency: 'Day'
    startTime: '2024-01-01T19:00:00+01:00'  // 7 PM
    timeZone: 'W. Europe Standard Time'
  }
}

3. Reserved Capacity (Premium)​

CommitmentDiscount
1-year~30-40%
3-year~50-60%

💡 Contact Microsoft for reserved instance pricing on Premium tier

4. Optimize Caching​

<!-- Cache responses to reduce backend calls (and costs) -->
<inbound>
    <cache-lookup vary-by-developer="false" 
                  vary-by-developer-groups="false"
                  downstream-caching-type="public"
                  caching-type="internal" />
</inbound>
<outbound>
    <cache-store duration="3600" />
</outbound>

5. Rate Limiting to Control Costs​

<!-- Prevent cost overruns from excessive usage -->
<rate-limit-by-key 
    calls="10000" 
    renewal-period="3600"
    counter-key="@(context.Subscription.Id)" />

📉 Cost Monitoring​

Azure Cost Management Query​

// APIM costs by resource
AzureDiagnostics
| where ResourceProvider == "MICROSOFT.APIMANAGEMENT"
| summarize Cost = sum(CostUSD) by Resource, bin(TimeGenerated, 1d)
| render timechart

Budget Alerts​

resource budget 'Microsoft.Consumption/budgets@2023-11-01' = {
  name: 'apim-monthly-budget'
  properties: {
    amount: 5000
    category: 'Cost'
    timeGrain: 'Monthly'
    timePeriod: {
      startDate: '2024-01-01'
    }
    filter: {
      dimensions: {
        name: 'ResourceGroupName'
        values: ['rg-apim-prod']
      }
    }
    notifications: {
      Actual_80: {
        enabled: true
        operator: 'GreaterThanOrEqualTo'
        threshold: 80
        contactEmails: ['platform-team@company.com']
      }
      Actual_100: {
        enabled: true
        operator: 'GreaterThanOrEqualTo'
        threshold: 100
        contactEmails: ['platform-team@company.com']
      }
    }
  }
}

🏷️ Cost Allocation Tags​

resource apim 'Microsoft.ApiManagement/service@2023-05-01-preview' = {
  name: apimName
  location: location
  tags: {
    CostCenter: 'IT-Platform'
    Environment: environment
    Project: 'DataMesh'
    Owner: 'platform-team@company.com'
    ChargebackProduct: 'API-Gateway'
  }
  // ... rest of config
}

Tag-Based Cost Report​

TagPurpose
CostCenterDepartment chargeback
EnvironmentDev/Test/Prod split
ProjectProject allocation
OwnerContact for review
ChargebackProductProduct-level cost

📊 Cost Comparison: v1 vs v2 Tiers​

FactorStandard (v1)Standard v2
Base price~$700/mo~$280/mo
Scale unitIncluded~$70/unit/mo
VNet supportNoYes
Zone redundancyNoOptional
Min deployment1 unit1 unit
Max scaleLimited10 units

When v2 is More Cost-Effective​

Standard v1 = $700/month fixed

Standard v2 at different scales:
- 1 unit  = $280 + $70  = $350/month  ✅ Cheaper
- 3 units = $280 + $210 = $490/month  ✅ Cheaper
- 5 units = $280 + $350 = $630/month  ✅ Cheaper
- 6 units = $280 + $420 = $700/month  ≈ Equal
- 7+ units = More expensive than v1

💡 v2 is more cost-effective up to ~6 scale units

🔄 Autoscaling for Cost Efficiency​

Scale Based on Demand​

resource autoscale 'Microsoft.Insights/autoscalesettings@2022-10-01' = {
  name: 'apim-autoscale'
  location: location
  properties: {
    enabled: true
    targetResourceUri: apim.id
    profiles: [
      {
        name: 'Business Hours'
        capacity: {
          default: '2'
          minimum: '2'
          maximum: '5'
        }
        recurrence: {
          frequency: 'Week'
          schedule: {
            days: ['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday']
            hours: [8]
            minutes: [0]
            timeZone: 'W. Europe Standard Time'
          }
        }
        rules: [
          {
            metricTrigger: {
              metricName: 'Capacity'
              metricResourceUri: apim.id
              timeGrain: 'PT1M'
              statistic: 'Average'
              timeWindow: 'PT10M'
              timeAggregation: 'Average'
              operator: 'GreaterThan'
              threshold: 70
            }
            scaleAction: {
              direction: 'Increase'
              type: 'ChangeCount'
              value: '1'
              cooldown: 'PT20M'
            }
          }
        ]
      }
      {
        name: 'Off Hours'
        capacity: {
          default: '1'
          minimum: '1'
          maximum: '2'
        }
        recurrence: {
          frequency: 'Week'
          schedule: {
            days: ['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday']
            hours: [19]
            minutes: [0]
            timeZone: 'W. Europe Standard Time'
          }
        }
        rules: []
      }
    ]
  }
}

✅ Cost Optimization Checklist​

Planning​

  • Right tier selected for workload
  • v2 tiers evaluated for VNet needs
  • Reserved capacity evaluated (Premium)
  • Dev/Test uses Developer tier

Operations​

  • Caching implemented
  • Rate limiting configured
  • Autoscaling enabled
  • Off-hours scaling configured

Monitoring​

  • Cost tags applied
  • Budget alerts configured
  • Monthly cost reviews scheduled
  • Cost anomaly detection enabled

Optimization​

  • Unused APIs removed
  • Cache hit rates monitored
  • Scale unit utilization tracked
  • Tier right-sizing reviewed quarterly

🔗 Related Documents​

DocumentDescription
01-ArchitectureTier comparison
04-PoliciesCaching and rate limiting
06-MonitoringMetrics for optimization

Next: 10-Performance-Efficiency - Caching, autoscaling, and latency optimization

📖Learn