AP-006 — Cost effective¶
Infrastructure costs are controlled by design, not by policy. The system is built to be economical at every scale.
Consumption-based infrastructure is preferred where workloads are variable. Right-sized compute. Caching used deliberately. Over-engineering for volume that does not yet exist violates both this principle and KISS.
The cost curve test: What does this cost per thousand transactions at launch? What at a million customers? If the curve is exponential, redesign before you build.
Cost tagging on all cloud resources. Monthly cost reviews in the engineering cadence. Nobody discovers a large unexpected bill at end of month.
Relationship to other principles¶
| Principle | Relationship |
|---|---|
| AP-001 KISS | Over-engineering for scale that does not yet exist violates both AP-001 and AP-006. |
| AP-007 Evolution by design | Right-sized now with a tested cost curve is more valuable than over-built for hypothetical future scale. |
| AP-008 Real time where it matters | Real-time infrastructure has a significant cost premium — it must be justified by a specific use case. |
| AP-009 Robust and serviceable | Operational complexity has a direct cost in engineering time — simpler systems are cheaper to run. |
See the full architectural principles index.