Key Trends and Insights in Architecture and Design (2019)

Key Points

There is growing demand for "evolutionary architecture" designs, building on previous discussions of replaceability and the importance of "glue" components in architecture; evolutionary architecture supports future changes in functional and cross‑functional requirements and constraints.

The awareness of the "microservices" style may be entering the late‑majority phase, but topics related to "correctly designing distributed systems", as well as reactive and fault‑tolerant design, are not following the same adoption curve.

We hypothesize that some architectural topics will never move to the early‑ or late‑majority phases; unfortunately, these include efficient patterns for specific use‑cases such as event‑sourcing/CQRS or Actor‑model systems.

The role of the "architect" is increasingly seen as one focused on technical leadership, pattern recognition, framework awareness, and designing solutions that span domains.

Although the term "serverless" can be vague, we appreciate its potential to emphasize event‑driven system design and the automatic elimination of certain platform concerns when implemented correctly.

InfoQ and QCon focus on topics we consider to be in the "innovators, early adopters, and early majority" stages. We aim to identify ideas that fit Geoffrey Moore's early market, where the audience consists of tech enthusiasts and visionaries seeking to stay ahead of emerging opportunities or pressing problems, and we look for concepts that might "cross the chasm" for broader adoption.

This article outlines our current view of the Architecture & Design (A&D) space, emphasizing infrastructure patterns, implementation of framework patterns, and the design processes and skills software architects need to cultivate.

Since our last review, notable changes include microservices moving into the late majority, while discussions continue to stress that designing correct distributed systems, reactive and fault‑tolerant designs, do not follow a straightforward adoption curve. We may be approaching the bottom of the "microservices hype" trough.

We also speculate that certain architectural domains will never progress beyond the early adopter phase; these include efficient patterns such as event‑sourcing/CQRS or Actor‑model systems, which may remain niche solutions for specific organizational problems.

While we find the term "serverless" somewhat ambiguous, we value its focus on modular, event‑driven system design and the automation of underlying platform concerns. Relatedly, we see increasing discussion around evolutionary architecture that supports future demand and constraint changes.

We are observing that the "architect" role is becoming more centered on soft skills like technical leadership, in addition to modern technical abilities such as pattern recognition, framework awareness, and cross‑domain design.

For context, this is the topic map from the second half of 2018; the 2019 version appears at the top of the article.

Below is a lightly edited excerpt from internal chat logs among three InfoQ Architecture & Design editors, providing additional context for our positioning on the adoption graph.

Daniel Bryant, independent tech consultant, Datawire product architect, and InfoQ news manager: As a ten‑person launch, I think HTTP/2 is moving toward early adopters (EA), HTTP/3 into innovators. GraphQL (and possibly gRPC) might be EA (or innovators?). I also think Chaos Engineering belongs in the DevOps queue. Microservices could be late majority (LM), and BDD, DDD, TDD as well. I would love to see "evolutionary architecture" appear somewhere – maybe EA? And what about "architects as technical leaders" (emphasizing the non‑technical evolution of the role)? I’m interested in hearing your thoughts and whether we need to move, add, or remove topics.

Jan Stenberg, IT consultant with 25+ years building systems on .Net/C# and JVM/Java platforms: I think Architecture & Design (A&D) is somewhat distinct from other topics we discuss on InfoQ. In A&D we don’t have a regular baseline of new or updated architecture versions. Instead, new tools, frameworks, or intelligent architectures make existing ideas resurface, possibly repackaged and branded. There are areas that could fit both queues. At a higher level they may belong to A&D, while the more technical side fits another queue. I think serverless is an example – at a higher level it’s an important A&D area, while the technical side may belong to the cloud queue. Micro‑frontends and similar technologies are another example – are they A&D or HTML5/JavaScript? There are also some areas I believe will never reach EM or LM stages, unfortunately they include some of my favorite architectures such as event‑sourcing/CQRS or Actor‑model systems. I suspect they will remain niche for the foreseeable future. So, my view (or hope) for the future of A&D:

Serverless – I hear it will become more automated, reducing underlying infrastructure workload.

Workflow platforms like Camunda – important for complex business logic in microservices or distributed systems.

Event sourcing/CQRS – I hope it becomes more mainstream, possibly EA or EM.

Event‑driven architecture – EA or EM.

Actor model/reactive – I discussed this with Vaughn Vernon last year; he believes it may become mainstream, though I remain skeptical.

Evolutionary architecture – interesting, I think EA is appropriate.

Chaos Engineering – usually part of DevOps; discussing it from an A&D perspective could be an exception.

GraphQL and similar tools – I see them as I or EA, potentially replacing REST (if correctly implemented).

Architects as technical leaders – many architects I meet focus on helping business/government domain experts understand their own fields; perhaps this aligns more with an agile story.

Microservices – LM. (I think microservices will soon become "today’s SOA" and are being used well, though many over‑label distributed systems.)

DDD – late majority, but I hope it remains an interesting InfoQ topic.

BDD – a late or "late minority".

TDD – still has some interesting discussion, whether too much or too little, unit vs. black‑box testing, etc.; likely LM at best.

When I encounter architects, developers, domain experts, and others in daily life rather than meetings, I often realize many of the concepts we discuss are unknown or very scattered for them, which also shows the value of InfoQ. I recall a talk at a developer conference in Canada where Vaughn Vernon asked how many people knew DDD; about half raised their hands. When I started writing for InfoQ, I wrote some updates about frameworks and libraries that might affect architecture, but over time my writing focused more on interesting blog posts and presentations, with only a few projects related to Axon, Akka, and other frameworks tightly coupled to specific architectures. Having this discussion at QCon was great.

InfoQ Editor‑in‑Chief Charles Humble: Vaughn Vernon and I discussed the Actor model – I think it may become mainstream, either directly or via messaging. In the JVM world, Akka does a good job promoting this approach; message‑driven systems have long been popular in financial systems as an Actor‑like method. Actors seem easy to grasp and reason about, and are a good way to handle large‑scale parallel work. I’d like to see momentum around Pony as a modern, Actor‑based system, though I doubt it will happen. Regarding evolutionary architecture, I’m interested in hearing Martin Fowler’s recent podcast on the topic and his involvement with extreme programming. I look forward to reading his book on ThoughtWorks.

Thomas Betts, IHS Markit Chief Software Engineer and InfoQ Architecture Queue Lead: At a high level, I agree with most of Daniel’s points. Jan is correct that some architecture patterns fit naturally into the topic‑graph progression, while others may never move beyond early adopters because they aren’t broadly applicable. I sometimes get confused about the overlap between A&D and other InfoQ topics, especially culture and methodology – perhaps Conway’s Law is to blame. Much of architecture boils down to communication – what are the external communication points into and out of the system? How will internal services talk to each other? How will data be stored and accessed? In many ways, a company’s answers and options will be based on where they sit on the A&D and C&M adoption‑lifecycle curve. I’d say A&D is the technical side of the equation, while C&M is non‑technical, though that’s an oversimplification. Technical implementation may fall into development or language queues. A&D sits in the weak spot between the two, handling cross‑cutting concerns and hoping to guide implementation. I’ll stop the philosophical roar and add some concrete discussion points.

Serverless – I personally dislike the term because it lacks a precise meaning; it may belong in EA.

Reactive – possibly EA. I think reactive architecture will become more common as developers get familiar with reactive programming, especially in JavaScript.

DDD – while DDD itself may move to LM, many derivative ideas are closely tied to it and could be in I or EA. Event sourcing, for example, could be EA/LM, though many sub‑topics may not fit the A&D topic map.

Microservices – using "serverless" as a frequently misused term. I think it is entering the late majority for broad adoption, though it may be EA for solid distributed architecture.

Distributed systems – I don’t think it’s a useful item on the topic map because it’s too broad. However, ideas like reactivity and fault‑tolerance are crucial for robust distributed systems, even if they don’t fit cleanly on the A&D map.

I fully support having this discussion at QCon!

The InfoQ editorial team is built by recruiting and training expert practitioners to write news and articles, analyze current and future trends, and invite editors to apply and join the conversation.

Original source: https://www.infoq.com/articles/architecture-trends-2019

Additional reference: http://pub.intelligentx.net/architecture-and-design-infoq-trends-report-january-2019

Discussion: Knowledge Planet 【Digitalization and Intelligent Transformation】