This post is part of a six-part series on Microservices Principles. Other parts are: Business Capability, Autonomy, Small bounded context, Asynchronous Communication and One Team.
Microservices are a hot topic. Because of that a lot of people are saying a lot of things. To help organizations make the best of this new architectural style Xebia has defined a set of principles that we feel should be applied when implementing a Microservice Architecture.
In this blog we cover: “Best technology for the job over one technology for all”
A common benefit of service based (and loosely coupled) architectures is the possibility to choose a different technology for each service. Even though this concept isn’t new, it’s rarely applied. Often the reason for this seems to be that even though the services should operate independently they do share (parts of) the same stack. This is further fueled by an urge to consolidate all development under a single technology. Reasoning here usually being that developers become more interchangeable and therefore more valuable if everything runs on the same technology, which should be a good thing.
So, if this isn’t new territory why drag it up again? Why would a Microservices architecture merit changing an existing approach? The short answer is autonomy. The long(er) answer is that a Microservices Architecture does not try to centralize common (technological) functions in singleton-esque services. No, the focus of a Microsservices architecture is on service autonomy, centered around business capability and a Microservice can therefore implement its own stack. This to make a Microservice easier to deploy on their own and removes dependencies on other services as much as possible.
But autonomous deployment isn’t the most important reason to consider technology on a per-service basis. The most important reason is the simple “use the best tool for the job”. Not all technology is created equal. This isn’t limited to the choice of programming language or even the framework. It applies to the whole stack including the data layer.
Instead of spending a significant sum to buy large, bloated, multipurpose middleware, consider lightweight, single purpose containers. Pick containers that run the tech you need. You don’t need Java applications with a relational database for everything. Other languages, frameworks and even datastores exist that cater to specific needs. Think of other languages like Scala or Go, frameworks like Akka or Play and database alternatives that focus on specific needs like storing (and retrieving) geographical data.
The choice of stack also relates to the choices you can make for your application landscape. If you have existing components that work for you or if you have components you want to buy off the shelve, it’s a real benefit to not be limited by an existing stack. For example, if you have opted for a Windows-only environment you are limiting your options.
Concerns about maintaining such a diverse landscape should consider that a lot of complexity comes from trying to maintain a single stack for everything. Smaller and simpler stacks should be easier to maintain. And having a single operations team for all those different technologies doesn’t sound like a good idea? You’re right! If you still have separate development and operations teams it may also be time to revisit that strategy. The devops approach makes running the services a shared responsibility. This doesn’t happen overnight but it is also a reason why Microservices can be such a good fit for organizations that have adopted an Agile way of working and/or apply Continuous Delivery.
Finally giving your developers a broader toolset to play with should keep them engaged. The opportunity to work with more than one technology can be a factor in retaining and attracting talent.
[edited: 3 aug 2015 – added preamble and removed line “Over the next couple of days we will cover each of these principles in more detail in a series of blog posts.“]