Beyond the Hype: When Microservices Beat (Modular) Monoliths in 2025
Abbas, Syed Ibtehaj (2025)
2025
Master's Programme in Computing Sciences and Electrical Engineering
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
Hyväksymispäivämäärä
2026-05-19
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202605185817
https://urn.fi/URN:NBN:fi:tuni-202605185817
Tiivistelmä
Over the past decade, microservices have become the default option at industry conferences. Agility and independent scaling were over that. By 2025, as will testify research and practice, this image will make it hard to see. Several organisations embraced microservice designs that proved ineffective when these organisations lacked the maturity of the platform or the team structure that could support them.
This study uses a non-systematic grey literature review covering 51 academic and industry sources published between 2020 and 2025, with over half drawn from 2024 and 2025. It addresses three research questions: (1) under what conditions does defaulting to microservices prove misguided, and when does a modular monolith perform better; (2) which organisational arrangements and platform capabilities must be in place for microservices to succeed; and (3) which workload traits and architectural properties give microservices a distinct advantage over a modular monolith.
The results indicate that team cognitive load, as developed by Skelton and Pais, should guide architectural choices rather than fashion. Microservices tend to succeed only when strong platform engineering supports them and when the organisation has a team structure, such as the Team Topologies model, which gives each team clear ownership of its services. Building on that finding, this thesis proposes a staged decision framework. A modular monolith fits teams that lack the capacity to manage the complexities of distributed systems. Microservices excel only when several conditions align: the domain changes at uneven rates across bounded contexts, selective scaling is required, data ownership boundaries are clear, and the organisation has the platform maturity to sustain distribution. The framework has four gates. Practitioners apply them when choosing an architecture based on their organisational capacity and workload characteristics.
This study uses a non-systematic grey literature review covering 51 academic and industry sources published between 2020 and 2025, with over half drawn from 2024 and 2025. It addresses three research questions: (1) under what conditions does defaulting to microservices prove misguided, and when does a modular monolith perform better; (2) which organisational arrangements and platform capabilities must be in place for microservices to succeed; and (3) which workload traits and architectural properties give microservices a distinct advantage over a modular monolith.
The results indicate that team cognitive load, as developed by Skelton and Pais, should guide architectural choices rather than fashion. Microservices tend to succeed only when strong platform engineering supports them and when the organisation has a team structure, such as the Team Topologies model, which gives each team clear ownership of its services. Building on that finding, this thesis proposes a staged decision framework. A modular monolith fits teams that lack the capacity to manage the complexities of distributed systems. Microservices excel only when several conditions align: the domain changes at uneven rates across bounded contexts, selective scaling is required, data ownership boundaries are clear, and the organisation has the platform maturity to sustain distribution. The framework has four gates. Practitioners apply them when choosing an architecture based on their organisational capacity and workload characteristics.