Skip to main content

Cost Monitoring Implementation

Overview​

The Cost Monitoring system is now a single source of truth for all ecosystem costs, tracking expenses across all services in the Toto platform.

📊 Services Tracked​

✅ Currently Implemented​

  1. AI/ML Services (Google Gemini) - Real-time

    • Fetched from toto-ai-hub ModelSelectionService
    • Tracks: Token usage, model costs, cost savings
    • Status: ✅ Real-time data

  2. Firestore Database - Estimated

    • Calculated from document counts
    • Tracks: Reads, writes, storage costs
    • Status: ⚠️ Estimated (needs Cloud Billing API for accuracy)

  3. SendGrid Email - Estimated

    • Calculated from user count
    • Tracks: Email volume, plan costs
    • Status: ⚠️ Estimated (needs SendGrid API integration)

  4. Sentry Error Monitoring - Estimated

    • Calculated from user count
    • Tracks: Event volume, plan costs
    • Status: ⚠️ Estimated (needs Sentry API integration)

  5. Twitter API - Fixed

    • Fixed monthly cost ($100 for Basic plan)
    • Status: ✅ Fixed cost

⚠️ Placeholders (Need Implementation)​

  1. Firebase App Hosting (Cloud Run)

    • Status: ⚠️ Placeholder (needs Cloud Billing API)
    • Projects: toto-app, toto-bo, toto-ai-hub

  2. Firebase Storage

    • Status: ⚠️ Placeholder (needs Storage usage metrics)

  3. Stripe Payment Processing

    • Status: ⚠️ Not included (transaction fees, not direct costs)

  4. Vertex AI Vector Search

    • Status: ⚠️ Not included (optional service)

🏗 Architecture​

API Endpoints​

  1. /api/cost-monitoring - AI-only costs (legacy)
  2. /api/cost-monitoring/comprehensive - All ecosystem costs (NEW)

Data Flow​

toto-bo Dashboard
    ↓
/api/cost-monitoring/comprehensive
    ↓
┌─────────────────────────────────────┐
│  Cost Aggregation Service           │
├─────────────────────────────────────┤
│  • AI Costs (toto-ai-hub)          │
│  • Firestore (estimated)            │
│  • SendGrid (estimated)             │
│  • Sentry (estimated)               │
│  • Twitter API (fixed)              │
│  • App Hosting (placeholder)        │
│  • Storage (placeholder)             │
└─────────────────────────────────────┘
    ↓
Unified Cost Dashboard

📈 Current Capabilities​

What Works Now​

  • ✅ Real-time AI cost tracking
  • ✅ Estimated costs for Firestore, SendGrid, Sentry
  • ✅ Fixed costs for Twitter API
  • ✅ Category-based cost breakdown
  • ✅ Service-level cost tracking
  • ✅ Toggle between AI-only and comprehensive view

What Needs Enhancement​

  1. Cloud Billing API Integration

    • Get real Firestore costs
    • Get real Cloud Run costs
    • Get real Storage costs
    • Priority: High

  2. Service API Integrations

    • SendGrid API for actual email usage
    • Sentry API for actual event counts
    • Priority: Medium

  3. Historical Data Storage

    • Store daily/weekly/monthly snapshots
    • Enable trend analysis
    • Priority: Medium

  4. Cost Alerts

    • Budget alerts
    • Anomaly detection
    • Priority: Low

🔧 Implementation Details​

Cost Calculation Methods​

Real-time (AI)​

  • Fetched directly from toto-ai-hub
  • Uses ModelSelectionService analytics
  • Most accurate

Estimated (Firestore, SendGrid, Sentry)​

  • Based on document counts and user activity
  • Uses pricing formulas from cost projections
  • Reasonable estimates for planning

Fixed (Twitter API)​

  • Known monthly subscription cost
  • No calculation needed

Pricing Formulas​

All pricing constants are defined in /api/cost-monitoring/comprehensive/route.ts:

const PRICING = {
  cloudRun: { ... },
  firestore: { ... },
  storage: { ... },
  sendgrid: { ... },
  sentry: { ... },
  twitter: { ... },
};

🚀 Next Steps for Full Accuracy​

Phase 1: Cloud Billing API (High Priority)​

  1. Set up Google Cloud Billing API access
  2. Create service account with billing viewer role
  3. Implement cost export queries
  4. Update comprehensive endpoint to use real data

Estimated effort: 2-3 days

Phase 2: Service API Integrations (Medium Priority)​

  1. SendGrid API integration

    • Track actual email sends
    • Get plan usage

  2. Sentry API integration

    • Track actual event counts
    • Get plan usage

Estimated effort: 1-2 days per service

Phase 3: Historical Tracking (Medium Priority)​

  1. Create Firestore collection for cost snapshots
  2. Schedule daily cost snapshots
  3. Build trend analysis UI

Estimated effort: 2-3 days

📋 Usage​

Accessing the Dashboard​

  1. Navigate to /dashboard/cost-monitoring in toto-bo
  2. Toggle between "All Costs" and "AI Only" views
  3. Select time range (24h, 7d, 30d, all)
  4. View comprehensive breakdown by service and category

Understanding the Data​

  • Real-time: Data fetched live from services (AI costs)
  • Estimated: Calculated from usage metrics (Firestore, SendGrid, Sentry)
  • Fixed: Known subscription costs (Twitter API)
  • Placeholder: Not yet implemented (App Hosting, Storage)

⚠️ Important Notes​

  1. Estimated Costs: Some costs are estimated and should be used for planning, not exact billing
  2. Data Freshness: AI costs are real-time; other costs refresh every 5 minutes
  3. Accuracy: For production budgeting, integrate Cloud Billing API for accurate costs
  4. Missing Services: App Hosting and Storage costs require Cloud Billing API integration

🔗 Related Documentation​

Last Updated: January 2025
Status: ✅ Comprehensive view implemented, Cloud Billing API integration pending