9 - Conclusions

We are satisfied with the platform developed, which originated from our business idea. The main achievements are:

• Achievements of project goals. All project goals were reached, with particular focus on aspects defining a distributed system. The system consists of isolated, loosely coupled components that appear to users as a single coherent platform.
• Requirements fulfillment. All functional and non-functional requirements were satisfied, with iterative refinement based on feedback from potential users.
• Microservices architecture design. The system was designed as a collection of independent services with clear boundaries. Backend microservices communicate and coordinate via a message broker and expose well-defined REST APIs used by the frontend.
• Distributed architecture implementation. Services were containerized using Docker and deployed both with Docker Compose and Docker Swarm, simulating a distributed environment with service isolation and internal networking.
• Modularity and extensibility. The architecture provides a clear separation of concerns, ensuring maintainability, easy addition of new features and future scalability. The modular design also facilitated parallel development of different functionalities.

9.1 - Future Works

Although the platform was designed according to distributed principles, it was deployed on a single physical machine. As a result, the system does not provide real high availability: if the host machine fails, all services become unavailable. Future improvements could include deployment on multiple physical or virtual machines to achieve infrastructure-level fault tolerance and resilience.

From a functional perspective, potential extensions include:

• Allowing organizations to respond to reviews.
• Additional notifications, such as friendship suggestions and updates on liked events.
• Profile insights and statistics.
• Refund management.
• Multi-currency support to enhance internationalization.

9.2 - What we learned

This project allowed us to apply distributed system concepts that we had previously studied only from a theoretical perspective. In particular:

• We gained practical experience with the trade-offs between consistency and availability in the presence of network partitions, as described by the CAP theorem.

• We experimented with containerization and faced challenges in guaranteeing data consistency across multiple services and system correctness when multiple replicas are deployed.

• We gained a deeper understanding of service decoupling, transparency, reliability and resilience both in single and multiple node deployment.

• We learnt how to make a real online deployment using a proprietary domain and physical server.