Not every architecture decision deserves a week of debate. Here are the ones that do — and the ones you should just pick and move on.
Decisions Worth Debating
Monolith vs. microservices. This shapes your team structure, deployment complexity, and operational cost for years. Start monolith unless you have a clear, proven need to distribute. Database choice. Changing databases is painful. SQL Server, PostgreSQL, or a managed NoSQL store — pick based on your data patterns, not trends. Authentication strategy. Identity is foundational. Cookie-based auth with ASP.NET Identity is battle-tested for most web apps. Don't roll your own.Decisions to Just Make
Which CSS framework. Tailwind, Bootstrap, whatever — pick one and move. You can always change it later. ORM vs. raw SQL. Use EF Core. If you hit performance issues, drop to raw SQL for those specific queries. Don't pre-optimize. Folder structure. Feature-based organization works. Stop bikeshedding and start building.The Decision Record
Document the decisions that matter in Architecture Decision Records (ADRs). One page: context, decision, consequences. Your future self will thank you.
Key Takeaways
- Spend decision energy on things that are hard to change
- Default to the boring, proven choice
- Document the why, not just the what
- Ship something — perfect architecture doesn't exist