Two Pricing Models Living on the Same Diagram

Every node you drag onto the canvas settles into one of two cost models. Compute-shaped services — the default Postgres Primary (4 vCPU / 16GB RAM / 100GB storage), Redis Cache (4GB RAM), MongoDB (16 vCPU / 64GB RAM / 2TB storage), or Kafka (8 vCPU / 32GB RAM / 500GB storage) — carry vCpu, ramGB, and storageGB fields that feed the same calculateMonthlyCost(provider, vCpu, ramGB, storageGB) function used across the rest of this site, so switching the provider dropdown between AWS, GCP, and Azure recalculates every one of those nodes instantly. Managed and serverless services instead carry a flat fixedCost that doesn't scale with a CPU/RAM shape: API Gateway (~$350/mo), Datadog (~$1,500/mo), DynamoDB (~$200/mo), Snowflake (~$1,000/mo base), and ALB (~$40/mo) are all modeled this way because their real-world billing is request-count or feature-tier based, not instance-shaped.

What the Total Doesn't Capture

• Cross-node data transfer: The edges you draw between nodes are visual only — connecting a frontend to a database doesn't add inter-AZ or egress bandwidth cost to the total.
• Committed-use pricing: Every number is an on-demand list price; 1- or 3-year reserved/savings-plan discounts (typically 30-60% off compute) aren't modeled per-node, only through the global discount toggle.
• Fixed-cost nodes ignore the provider switch: Flipping AWS/GCP/Azure recalculates shape-based nodes but leaves fixed-cost services (Datadog, API Gateway, KMS) untouched, since those flat numbers aren't provider-specific in this model.