Architecture Principle: Data gravity creates financial gravity. Your baseline cost is dictated by data mass; your multipliers are dictated by routing.
This repository contains the Terraform decision matrices and cost surface models for optimizing egress routing.
The continuously maintained architectural specification lives here: 👉 https://www.rack2cloud.com/physics-of-data-egress/
Cloud cost overruns are frequently attributed to compute scale but originate from underestimated data egress patterns.
If your architecture uses a NAT Gateway for internal cloud service access (like AWS S3 or DynamoDB), you are paying for data transfer twice: once for the NAT processing fee ($0.045/GB), and once for the egress. This model maps cost as a function of data movement topology to eliminate double-metering.
Nodes (Billable Boundaries):
- Data Origin (VPC/Subnet)
- Inter-region traffic
- Cross-cloud transfer
- On-prem exit
- SaaS consumption
| Scenario | Recommendation |
|---|---|
| High S3 Throughput (VPC) | ✅ Use VPC Gateway Endpoint (Free) |
| Cross-Region S3 Access | ✅ Use VPC Interface Endpoint (PrivateLink) |
| Public API Access only | |
| Multi-Region Peering | ✅ Implement Hub-and-Spoke Transit Gateway |
By implementing VPC Gateway Endpoints, traffic to cloud-native storage (like S3) remains on the cloud provider's global backbone. It bypasses the NAT Gateway entirely, dropping the per-GB processing fee to $0.00.
Star this repo if you are actively optimizing cloud spend. Architectural frameworks maintained by Rack2Cloud.