In software development, starting with a "monolithic" structure is often both fast and sensible. But as the project grows, the combination of different teams working on the same codebase and a single point of failure that can bring the entire system down sooner or later forces an architectural transformation.
Designing Resilient Systems
If you're building a backend that handles tens of thousands of requests per second, it's not just code execution speed that matters—network and database response times become critical too. This is exactly where caching mechanisms (Redis), load balancing, and event-driven architectures come into play.
Isolating Components from Each Other
You don't want your services to work by waiting for each other (synchronous execution). By using message queue systems like Kafka or RabbitMQ, you can prevent one operation from blocking others. Once you set this up correctly, the pressure on the rest of the system decreases significantly—I've seen this firsthand in several projects.
Instead of a Conclusion
Architectural decisions are generally hard to reverse. A premature optimisation can hurt a project just as much as a late architecture overhaul; the real challenge is reading correctly what stage the project is in.