The folly of microservices

I’m sure I’m not the first to make some of these observations about the microservices fad, but here’s my perspective nonetheless.

Microservices are definitely all the rage, and proffered as an obvious solution to the dreaded monolith. But I'm not so sure it's all that obvious. As another swing at service-orientation, which is really about composability and modularity, it has the same appeal as previous iterations.

What I find disappointing about microservices is the emphasis on technologies over decomposition techniques. Its more concerned with how to run and deploy services than it is about how to identify the proper boundaries and granularity of services. It tells you how to run a service, without telling you what a service should be or do.

Even when it comes to technologies, it's not a no brainer. More services and more technologies (containers, orchestrators) often means more complexity to manage, and a system that is harder to reason about. That complexity management requires yet more technology. All fun and cool for us technology folks, but not clearly necessary.

Applying the technology solutions of microservices without the proper system decomposition will actually lead to some pretty bad outcomes (distributed monolith). There seem to be two class of problems here:

  • Taking non-modular code and distributing that over the network, is unlikely to improve anything.
  • Avoiding decomposition, by choosing extremely granular services, and distributing those over the network can be even worse.

Microservices are a technique to solve a certain type of problem.  They are not a goal in of themselves (unless you're doing resume-based development or a variant thereof). So it's important to understand what you are trying to solve with them.

My impression is that it is a solution to a problem of large teams, with well-defined responsibilities, that would benefit from the increased encapsulation, independence and deployment technologies. They do not seem like a good solution for a small team with broad responsibilities, or with poorly defined system boundaries and/or team responsibilities.

Sometimes modularity itself is enough, without all the tech, but where's the fun in that? And what's the next hiring manager going to say when I don't have years of microservice experience?

Comments

Post a Comment

Popular posts from this blog

RIF Notes #4

“It is better to be feared than loved, if you cannot be both” – Machiavelli “A well-stated problem is half-solved” – John Dewey jsFiddle - Cool javascript testing tool Modernizr - Modernizr is a small and simple JavaScript library that helps you take advantage of emerging web technologies (CSS3, HTML 5) while still maintaining a fine level of control over older browsers that may not yet support these new technologies. Why We’re Going With HTML(5) Instead Of Silverlight – More on the debate between Silverlight and HTML5.  A year ago I would have said Silverlight, hands down, for LOB development, now its not quite as clear.  At least not if your users need to access your apps from their iPads and Droids. Bootstrapped, Profitable, & Proud: Braintree – Braintree Payment Technologies was the first company I contacted when I first started investigating credit card tokenization a few years back.  Remote Desktop Protocol (RDP) -- Technical Analysis - Matt found this article, for an
Read more

RIF Notes #42

“In practice, problems are delegated but the power to address them is not.” - Maslach & Leiter Building the Object Model You Want with Entity Framework How you can improve your SQL with code analysis in SQL Prompt A growing number of young Americans are leaving desk jobs to farm – This is a head scratcher. Macro trends in the tech industry Technology Radar Vol 17 Essential .NET - C# 7.0: Tuples Explained – “What many folks didn’t realize when it was first introduced is that the new C# 7.0 tuple all but replaces anonymous types—and provides additional functionality.” Editor's Note – Misprint – “Any IT manager can regale you with stories of undiagnosable printer failures, but even at the home-office level printing is an exercise in fail” .NET Standard - Demystifying .NET Core and .NET Standard Devops - Continuous Data Migration Using Visual Studio and TFS
Read more

The greatest trick the AI ever pulled was convincing the world it wasn't THAT intelligent

Just listened to The AI Dilemma and it definitely elaborated on and articulated some of my own concerns, as did the news of OpenAI plugins. The synthesis of concern is: AI doesn't need to be that powerful to be destructive and destabilizing (e.g. social media algorithms) to people and society. These new LLM's are already much more powerful than social media algorithms. Many of the new LLM capabilities are emergent and poorly understood. The experts designing these systems don't fully understand them, and some of whom are already worried about what they've created. LLM's are being rapidly and recklessly deployed very broadly across society, in an arms race. The strongest argument against worry is that these LLM's aren't really intelligent, are really just a super-autocomplete that enables automating mundane tasks via natural language. But I don't think any of the companies developing or deploying these things are arguing that. Even if they were, see #1 &
Read more