Mental Models for Infrastructure Architects
Useful frameworks for thinking about complex infrastructure problems and making better design decisions.
1. First Principles Thinking
Strip the system down to fundamentals: what actually needs to happen, what doesn’t, and what’s just legacy inertia.
- Ask: What’s the real goal? What are we assuming, and why?
2. Inversion
Instead of asking “how do we scale this?” ask “how does this fail?”
- Avoiding failure is often more powerful than optimizing for success.
3. Second-Order Effects
Every change creates ripples. Always ask: “Then what?”
- What does this optimization break? What does this shortcut incentivize?
4. Chesterton’s Fence
If a piece of infra exists, assume it had a purpose—even if that purpose is no longer valid.
- Investigate before you remove. Preserve knowledge, not just code.
5. Local vs. Global Optimization
What looks efficient for one team may create systemic fragility.
- Design with the whole system’s feedback loops in mind.
6. Tradeoff Space Mapping
Every decision is a trade: latency vs. durability, cost vs. visibility, autonomy vs. control.
- Explicitly map tradeoffs. Accept that perfection is not possible.
7. Antifragility
Build systems that get stronger from stress, not just survive it.
- Encourage safe failure, fast detection, and graceful degradation.
8. Drift Accumulation
Systems always drift—configuration, understanding, dependencies.
- Invest in re-alignment: cleanup, re-audit, re-document.
9. Principle of Least Astonishment
Your infra should do what people expect. If it’s surprising, it’s dangerous.
- Clarity and predictability reduce operator error.
10. Feedback Loops
Healthy systems have visible, fast, honest feedback.
- Build observability into the core. No feedback = no learning.
Each of these models works better when held lightly, applied pragmatically, and revised based on what your systems actually show you—not what the diagrams promised.