Ovidiu Mățan: How do you see the difference between SOA (Service Oriented Architecture) and Microservices ?
Sam Newman: They are not different things. Microservices are an opinionated subset of SOA. SOA just describes a system built out of collaborating services. Microservices is a type of SOA, which is more opinionated - it's "prime directive" if you will is that microservices are independently deployable. So think of microservices as a type of SOA, in the same way that XP is a type of Agile.
Regarding the size of a microservice, you mentioned in your very known book: Building Microservices that it should be "as big as my head", based on James Lewis affirmation. What did you mean ?
Sam Newman: This version of "size" speaks to your ability to comprehend what the microservice is doing. It's just one measure you can use, but if you are unable to understand what the microservice is doing, it might be too big! The problem of course is that people have "different sized heads" - which is to say someone familiar with the system may better be able to understand what a microservice is doing than someone brand new to the project. Likewise a more experienced developer might more quickly understand what a given microservice is doing, compared to someone newer to the industry. So it's far from a perfect measure.
Moving to a microservices approach is also changing the company's teams structures?
Sam Newman: Microservices work best where you can tie back ownership of a microservice to a specific team. So one team might own more than one microservice, but each microservice should be owned by exactly one team. At a basic level, you may need to change your organisation to move away from more collective ownership models (anyone can change anything) to stronger forms of ownership.
Going further though, microservices can help you move to organisational models like those outlined in the Team Topologies book - steam-aligned teams, with end-to-end responsibility for building and running an application.
Your affirmation "microservices are a poor choice for an organization primarily concerned with reducing costs" is something that some of the IT managers should be looking before going into today's trend with microservices. Please explain us why it will be more expansive.
Sam Newman: It's pretty simple. Firstly, you will likely need new infrastructure, and new tooling to manage that infrastructure. Secondly, learning how to build and manage a distributed system is hard. This will take time and energy. This will in turn reduce the amount of time and energy you can put into new features. So you'll either have to be satisfied with a slower delivery of features, or else having to try and add people to offset this problem. Microservices *might* reduce costs in the long term though through driving other efficiencies, but this is likely a long way off. If your primary motivation is cost reduction, I'd consider microservices to be a bad choice.
Microservices and startups - what's your recommendation ?
Sam Newman: Probably don't do it. At the start of a new project, we often don't know exactly what we want to build, nor do we know if it will be successful. We need to be able to rapidly experiment, and understand what capabilities we need to build. If we tried building for massive scale up front, pass:[\we'd\] end up front-loading a huge amount of work to prepare for load that may never come, while diverting effort away from more important activities, like understanding if anyone will want to actually use our product. Eric Ries tells the story of spending six months building a product that no one ever downloaded. He reflected that he could have put up a link on a web page that 404\’d when people clicked on it to see if there was any demand, spent six months on the beach instead, and learned just as much!
I expand a lot on this here.
What's coming next after microservices and we should pay attention?
Sam Newman: After? Who knows? Microservices arguably are just a continuation of a lot of old ideas. They're a rediscovery (for some) of modular programming (which we've been doing since at least the 1960s). I would hope that people who are deciding against microservices would consider building modular monolithic applications instead.
In terms of general trends, expect Kubernetes to continue to be in more places, but increasingly expect you not to notice. It's not developer friendly. Abstractions like FAAS will hopefully become more common, allowing us to hide away that detail.
Tell us few words about your presentation from The Developers conference: Hiding the lead Coupling, cohesion and microservices
Sam Newman: This talk is all about looking back at concepts from and 1960s and 1970s which related to how we organised our code. Microservices are a modular architecture, albeit one with the added complexity of being distributed systems. The upside though is that we can take a lot of prior art from structured programming and elsewhere and apply it in the context of microservices, to make sure we actually deliver on their promise. Even if you aren't building a microservice architecture, or have no interested in doing so, there should still be some concrete takeaways for all attendees.