Modularity is the practice of dividing a system into separate, independent parts to improve communication and decision-making during the design process.
While microservices offer benefits like fast delivery, they can create a complex, tangled landscape that slows down innovation when new business requirements touch multiple services.
Instead of relying on intuition, teams should measure modularity at the product and domain level using concrete metrics, similar to how code quality is measured.
The average component dependency (ACD) metric helps quantify the overall connectedness of a system by averaging the direct and indirect dependencies for each microservice.
Propagation cost (PC) is a key metric that measures the ripple effect of a change in one component across the entire system, directly impacting innovation speed.
The lack of cohesion metric helps determine if a group of microservices truly belongs together by analyzing shared concerns, such as data access patterns.
By calculating the coupling factor and tracking metrics on a dashboard, teams can monitor trends and make informed decisions about system architecture and team structure.
The modular monolith pattern offers a middle ground by enforcing strong boundaries between modules while keeping them in a single deployable unit, simplifying refactoring.
Adopting modularity requires a mindset change that extends beyond engineering to business decisions, experimentation with metrics, and sharing learnings with the community.
To count direct and indirect connections in complex systems, tools like Istio for service mesh data and distributed tracing can be used to automate dependency discovery.
26:39Loïc Carbonne
30:12Hendrik Lösch
20:39Duan Lightfoot
24:25Florian Lenz
25:05Eberhard Wolff
43:13Carola Lilienthal
26:22David Tielke
24:38Carola Lilienthal
.gif?w=240&auto=compress,format)
.gif?w=240&auto=compress,format)
Jobs that call for the skills explored in this talk.
Vesterling Consulting GmbH
Frankfurt am Main, Germany
CONTIAMO GMBH
Berlin, Germany
