Tobias Keller·
Line-item AWS cost table that pinned CloudFront egress as my biggest non-discountable spend
Produces detailed cloud infrastructure cost estimates with modeling for growth scenarios, reserved capacity planning, and cost optimization recommendations.
Infrastructure Cost Estimation & TCO Analysis
You are a Cloud Financial Operations (FinOps) Architect who has optimized cloud spend for Fortune 500 companies. Create a detailed infrastructure cost estimate.
**Architecture Overview**: {{architecture_overview}} (components: compute, database, storage, CDN, messaging, caching, etc.)
**Cloud Provider**: {{cloud_provider}} (AWS/Azure/GCP with specific services planned)
**Traffic Projections**: {{traffic_projections}} (monthly active users, requests per second, data transfer volumes, storage growth)
**Optimization Constraints**: {{optimization_constraints}} (reserved instance commitment willingness, spot instance tolerance, budget ceiling)
Provide the complete cost analysis:
1. **Component-Level Cost Breakdown** - Line-item estimate for every infrastructure component with instance types, quantities, and monthly costs
2. **Pricing Models Comparison** - On-Demand vs Reserved Instances (1yr/3yr) vs Savings Plans vs Spot for each compute component
3. **Growth Modeling** - Cost at 1x, 5x, 10x, 50x scale with cost-per-user metrics and economies of scale identification
4. **Data Transfer Cost Analysis** - Cross-AZ, cross-region, internet egress, NAT Gateway costs with optimization strategies
5. **Storage Tiering Costs** - Hot/warm/cold/archive storage costs with lifecycle policy recommendations and retrieval cost modeling
6. **Database Cost Optimization** - Read replica vs cache cost comparison, serverless vs provisioned, storage IOPS optimization
7. **Third-Party SaaS Costs** - Monitoring (Datadog/NewRelic), logging, CI/CD, security tools, auth provider estimated costs
8. **Personnel Cost Factor** - DevOps/SRE headcount required for the architecture at each scale tier
9. **TCO Comparison** - Cloud vs on-premises vs hybrid over 3-year period with NPV calculation
10. **Cost Optimization Roadmap** - Immediate (0-30 days), Short-term (1-3 months), Long-term (3-12 months) optimization actions with savings estimates
11. **Budget Alerting Strategy** - AWS Budgets/Azure Cost Management setup, anomaly detection, chargeback/showback model
12. **Architecture Cost Trade-offs** - Cheaper alternatives for each component with performance/availability impact quantified
13. **Monthly Recurring Cost Dashboard** - Summary table of all fixed monthly costs vs variable usage-based costs
Include actual pricing calculations based on current AWS/Azure/GCP pricing and a 12-month projection spreadsheet structure.
Ergebnisse
## Infrastructure cost estimate — AWS, 50k MAU web app
Component-level monthly breakdown at current scale, plus the levers that actually move the bill.
| Component | Config | On-Demand | With commitments |
|-----------|--------|-----------|------------------|
| ECS Fargate | 3 tasks, 1 vCPU / 2GB, 24x7 | $190 | $133 (1yr Savings Plan) |
| RDS Postgres | db.t4g.medium Multi-AZ | $240 | $168 (1yr RI) |
| ElastiCache | cache.t4g.small | $50 | $36 |
| S3 + CloudFront | 2TB storage, 8TB egress | $720 | $720 |
| NAT Gateway | 1x + 3TB processed | $130 | $130 |
| **Total** | | **~$1,430** | **~$1,187** |
**The egress trap:** CloudFront data transfer ($720) is the largest line and is *not* covered by Savings Plans. Biggest win is raising the cache hit ratio (more bytes served from edge, fewer from origin) and enabling Brotli — realistically 20-30% off that number.
**Growth modeling**
| Scale | MAU | Est. monthly | $/user |
|-------|-----|-------------|--------|
| 1x | 50k | $1,190 | $0.024 |
| 5x | 250k | $4,100 | $0.016 |
| 10x | 500k | $7,200 | $0.014 |
Cost-per-user *drops* with scale — fixed costs (Multi-AZ, NAT) amortize. **Roadmap:** 0-30d buy Savings Plans + RIs (-18%); 1-3mo add a single-NAT-per-AZ review and S3 lifecycle to Glacier; 3-12mo evaluate Aurora Serverless v2 if DB load stays bursty. Budget alerts at 80% / 100% with anomaly detection on daily spend.
Modell: Claude Opus 4
35 Likes13 SavesScore: 20
3 Kommentare
Lena Fischer·
Pasted, tweaked two lines, shipped. Love it.
Jonas Weber·
This is the kind of prompt that pays for the subscription by itself.
Ahmed Hassan·
The system design details here are spot on.