system-design

Production System Design

Use this skill to turn business requirements and operational constraints into the simplest production architecture that can be built, operated, evolved, and recovered safely.

This is not an interview-diagram generator. Every component and complexity must earn its place through a requirement, measured constraint, or explicit risk.

System Design Contract

• Start with requirements, constraints, scale, and failure tolerance before selecting technologies.
• Separate facts, assumptions, estimates, and decisions.
• Quantify demand and SLOs enough to expose bottlenecks; never invent precision.
• Tie every consequential mechanism to a quality-attribute scenario, measurable evidence, and a rollback or revisit trigger.
• Prefer a modular monolith or one ownership boundary until independent scaling, deployment, security, data ownership, or team autonomy justifies a split.
• Keep data ownership explicit. Shared databases and synchronous call chains create coupling even when diagrams show separate services.
• Map correlated failure domains and hidden shared dependencies; replicated components are not isolated when they share a control plane, data store, quota, identity service, or deployment path.
• State consistency and durability guarantees at each boundary.
• Design degraded behavior, recovery, and migration, not only the healthy steady state.
• Treat security, privacy, operability, and cost as design inputs, not review-stage additions.
• Verify current vendor limits, guarantees, defaults, and pricing against official sources before relying on them.

Workflow

1. Frame the problem

Capture:

• users, business outcome, and critical journeys;
• functional requirements and explicit non-goals;
• compliance, privacy, residency, and trust boundaries;
• existing system and migration constraints;
• delivery timeline, team ownership, and operational capability.

Ask at most three focused questions that would materially change the architecture. When answers are unavailable, proceed with clearly labeled assumptions and explain their impact.