MODEL OF WEBAPPLICATION ARCHITECTURE BASED ON MODERN MULTILAYERED ARCHITECTURES

Abstract

The paper explores the relevant task of updating classic multi-layered web application architectures in view of the increasing requirements for development speed, flexibility and scalability of modern software solutions. The authors have conducted a thorough analysis of traditional approaches (N-layered, Onion, Clean Architecture) and identified their key limitations in the context of dynamic project development. Attention is paid to the problem of "vertical dependency", in which changes in one business function require code adjustments at all horizontal levels, which causes excessive coherence and complexity of the system logic. An improved web application architecture model is proposed, which covers the presentation, application, domain and infrastructure levels, cross-components, specialized validation mechanisms and integration of modern architectural patterns (CA, VSA, CQRS). The paper describes in detail the content of each level: from the use of SSR and CSR in the presentation layer to the use of message brokers and caching in the infrastructure. The central place in the proposed model is occupied by the implementation of the principles of Domain-Driven Design (DDD) and the Rich Domain Model approach, which ensures the encapsulation of business rules directly in domain entities and prevents the occurrence of incorrect system states. The concept of vertical slices is also considered, within which each business function is implemented as an autonomous module, which reduces the interdependence between components. The practical effectiveness of the proposed model is confirmed by several examples. In particular, the use of a hybrid approach made it possible to reduce the infrastructure deployment time from three weeks to one. The development of new functions is accelerated by up to 60%, and changes to existing code are made 70% faster compared to traditional architectural models. In addition, thanks to the use of the CQRS principle, it was possible to increase the performance of high load read operations by up to 50%, reducing the response time from 400 ms to 150-200 ms. Therefore, the proposed model retains the advantages of the classic multi-layer approach: structured code, ease of unit testing, and ease of scaling, while providing greater adaptability and flexibility of the architecture in the face of modern requirements