Coupling and Control Flow in Distributed Systems

Architect love to dissect buzzwords and try to keep people from tossing around technical terms with little understanding. One such term is “coupling” – it’s almost guaranteed to be mentioned when discussing large system design. It’s widely agreed that excessive coupling hampers software evolution and can make systems brittle. But how much coupling is “right”? How do you measure coupling in the first place? And how do you reduce it? Those nuances tend to get less airtime.

So, I decided to revive my Enterprise Integration Patterns blog and write an article about the implications of Event-Driven Architectures and coupling. I wanted to “myth-bust” the common slogan “Events make your system loosely coupled”.

I soon found out that a nuanced Architect Elevator-style discussion, requires a mental model for coupling first. That post alone grew to over 3,000 words. Soon after, I also realized the importance of distributed control flow in this context, e.g., whether a system pushes or pulls messages. So, the idea now turned into a growing series of blog posts on distributed system design, all on my Enterprise Integration Patterns site:

At over 10,000 words, sounds like the start or another book? Could be!

BTW, the Enterprise Integration Patterns site has been discussing distributed system design for over 20 years now. It is still a static site (ahead of the trend!) rendered by a combination of Apache Ant and XSLT (perhaps no longer ahead of the trend). Some topics and technologies do pass the test of time!