7 DevOps trends to watch in 2017
As IT organizations continue to implement or modernize their DevOps practices, it’s important not to get left behind. IT operations management and development groups need to paget a sense of where technologies are headed so they are ready to adapt when the time comes—or avoid the hype in some cases.
To help you understand the challenges and opportunities likely to arrive in 2017, TechBeacon spoke with experts in DevOps, cloud, microservices, containers, and emerging ecosystems such as serverless computing. Here’s what they had to say.
A clear definition of DevOps will finally emerge
IT companies have been struggling with DevOps transformations for years. J. Paul Reed, managing partner at Release Engineering Approaches, thinks that at least one struggle will end in 2017—the struggle to understand the exact definition of DevOps.
“2017 will be the year that DevOps is finally declared '1.0-stable'." —J. Paul Reed
It's no longer an emergent phenomenon because there's been a lot of work to define and codify DevOps as a static set of principles and practices.”
Reed discussed the dilemma of DevOps divergence in his argument that DevOps is disintegrating. While that divergence will continue, the generally accepted definition of DevOps will be promoted, while others become increasingly minimized. “As for what this ultimately means for DevOps, ask me in 2018,” he says. "We'll start to see this mechanism happen in 2017.”
Jeremy Likness, a blogger and director of application development at managed services provider iVision, has an idea of what that final definition will look like: the new application lifecycle management (ALM) methodology.
“Many organizations will challenge agile and recognize DevOps as the new ALM methodology that is a generation beyond agile, rather than a superset. As part of this shift, we'll see Infrastructure as Code (IoC) continue to gain a foothold in continuous delivery pipelines.” —Jeremy Likness
Testers will learn to code or perish
TJ Maher, an automation developer and TechBeacon contributor, spent the last two years updating his skills to move from manual tester to automation developer to software engineer in test. In those same two years, he’s seen many of his former QA testing colleagues lose their jobs due to the major changes going on in the testing industry right now.
“Continuous integration and continuous delivery turned the big splash of Selenium WebDriver into a tsunami that washed away almost all of the software testing industry, drowning many of the manual testers and eroding their base of employability.” —TJ Maher
For many testing engineers, 2016's motto was "learn to code or perish." Testing is now focused at the web services level, with tremendous demand for RESTful APIs, and Selenium wrappers, he says.
Andy Tinkham, global practice lead for QA testing at C2 IT Solutions Consulting, believes there’s another big reason why generic QA professionals are having trouble finding jobs: the commoditization of testers.
Entire careers were built on quality analysts moving from industry to industry, applying their knowledge of good testing to systems, he contends.
“We focused on making testing repeatable enough that anyone could do it. When combined with an automation approach that tries to directly translate human activity into scripts and a development culture that is blurring the lines between roles, we ended up commoditizing the testing industry.” —Andy Tinkham
As a result, Tinkham says, more tests have been automated (although he admits this is no panacea), and other roles have taken on the tester’s duties. And in some cases, testing has been transferred to offshore teams.
Tinkham and Maher both say 2017 will be a pivotal year for testers, with most jkobs requiring a higher degree of specialization than ever before. “Whether it’s a focus on data warehousing and ETL, a focus on automation, or a focus on some other aspect of testing that has previously been considered just one skill,” Tinkham says.
"Back to basics" agile movements will gain steam
15 years after the creation of the Agile Manifesto, agile and scrum are considered best practices by many, but others there have become jaded by the unintended side effects of dogmatic agile methodologies. “We’ve shifted away from those core values that made the Agile Manifesto so revolutionary in the first place,” says Tinkham, and this will be the year that “back to basics” agile movements gain steam.
Tinkham believes that two movements have the best chance to gain mainstream attention around the industry. One of them is Joshua Kerievsky’s Modern Agile, which was introduced in the Agile 2016 keynotes. The other is Heart of Agile, by Agile Manifesto signatory, Alistair Cockburn.
These philosophies have already been presented at agile conferences and are reinvigorating many who have previously become jaded. "Expect to see them grow in influence and begin to impact mainstream agile thinking over the next 12-18 months,” Tinkham says.
More organizations will move to cloud, but pass on PaaS
David Linthicum, senior vice president at Cloud Technology Partners and regular TechBeacon contributor, says 2017 will be the year that most organizations move to the cloud in a big way.
“If they moved 20 applications in 2016, then in 2017 they will move 500.” —David Linthicum
The way that companies use the cloud will also start to change this year. “Platform-as-a-Service (PaaS) will start to die a slow death... because it tightly couples the solution to the cloud vendor,” says Likness. Companies will instead favor container-based solutions that give them flexibility and portability in a hybrid environment that may contain services from multiple cloud vendors.
Expect to see more Linux commodity servers and organizations leveraging .NET Core to move away from dependency on Windows-based machines, he adds. "More vendors will provide not only software-as-a-service (SaaS), but containers-as-a-service, so customers running on-premise have easy access to an up and running, out-of-the-box solution.”
Microservices hype will begin to cool
The enthusiasm surrounding microservices was at an all-time high in 2016. While microservices are a great advancement for many applications, they’re not a magic bullet, says Marco Troisi, senior software engineer at Bluefin Payment Systems. Both he and Paul Bakker, software architect at Netflix, think that the over-hyping of microservices will die down in this year.
“There are going to be more people talking about when it's not a good idea to build software with a microservices architecture. At the same time, tools that help us manage a distributed architecture are going to reach a higher level of maturity, making it easier than ever before to work with microservices.” —Marco Troisi
Bakker says that many enterprises treat microservices as a synonym for modern, lightweight frameworks. “Of course these lightweight alternatives are a great way forward. But that doesn’t necessarily mean you need [a] distributed architecture as well,” he says. “For those who do not understand the distinction between architecture and tools, microservices will become the new service-oriented architecture (SOA) in 2017 and those firms will likely invest a lot of money in commercial tools they don’t actually need.”
Christian Posta, a principal solutions architect in Red Hat’s middleware specialist group, expects that many developers will make mistakes this year with regard to microservices.
“Enterprises will begin realizing that Java's not ideal for microservices developments; yet many of them will continue to invest in that direction." —Christian Posta
Posta said he believes that major Internet-based businesses such as Netflix, Twitter, and others, will rethink their open source strategies. Many of them, he says, are just “slinging code over the wall.”
Those changes will be important, he says, because those open source tools and libraries from major tech firms will start displacing traditional vendor middleware. “Some new interesting startups will spring up around this emerging ecosystem,” he says.
Containers and orchestration tools will become easier to use
A huge amount of investment from major cloud providers is going into containers. Container cluster management is a key area where providers are building solutions, says Likness.
Troisi expects toolmakers to focus on making Docker and other containers easier to use this year, and he's particularly excited about Docker Compose, which should be production-ready in 2017.
“Storing Docker commands in Compose's easy-to-read YAML files is going to become the preferred way for developers to run Docker apps, as opposed to having to remember huge, unreadable command-line interface commands,” says Troisi.
Linthicum expects general interest in containers to grow this year, but only around greenfield applications.
“It will be too hard and costly to containerize most older applications." —Jeremy Likness
Likness says containers became part of many development workflows last year. This year, he says, they will become just as prominent in production workflows.
Kubernetes will be the primary container orchestration engine
Kubernetes will be the de facto industry standard for container orchestration in 2017, Troisi predicts, and research seems to support this claim. But Kubernetes is still relatively difficult to set up and use, so container-based PaaS systems, such as RedHat's OpenShift and CoreOS Tectonic, will help ease IT organizations into the world of Kubernetes and container orchestration.
Bakker agrees with Likness’ “slow death” prediction for PaaS, but only for older, strict-pattern lock-in offerings, such as Google AppEngine. He expects PaaS offerings based on Kubernetes, such as Google Container Engine and OpenShift, to thrive in 2017. Posta expects the same: “Kubernetes is eating containers, and will continue to roll on; PaaS providers built on Kubernetes will gain more traction than other ones.”
The race between cloud providers is not about virtual machines anymore, says Bakker. Instead, the race will focus on container platforms, and making it as easy as possible to run containers in the cloud.
"2017 will be the year where hosted container platforms become what IaaS was a few years ago.” —Paul Bakker
The hype level for serverless will rival that of microservices and containers
Serverless (also known as Functions-as-a-Service or “FaaS”) is one of the newest trends in IT, and has massive potential to fundamentally change how some organizations develop software. Bernard Golden, CEO at Navica, author of "Amazon Web Services for Dummies,” and regular TechBeacon contributor, expects increased awareness and more early adoption of serverless technologies.
“Serverless holds the potential for IT organizations to get out of the infrastructure management business completely, and focus on application development and deployment. While IT has always been a domain of constant change, next year will offer more opportunity and challenge to IT organizations than they have ever seen before.” —Bernard Golden
Dean Hallman, chief technology officer and principal data systems consultant at software consultancy Cloudbox Systems, has been researching serverless frameworks since the first ones arrived in the open source space. He sees serverless frameworks creating more wizard-like experiences and filling in the feature gaps between major FaaS providers (such as AWS Lambda, Azure Functions, and IBM OpenWisk) so that serverless apps can target any of these vendors' services from a single codebase.
Hallman also expects serverless to have a major impact on the evolution of DevOps. Developers will be more involved than ever in domains that were previously managed by Ops and DevOps. “Most serverless frameworks already propose a serverless-friendly DevOps workflow,” Hallman says. “AWS Lambda also extended its platform to promote ‘blast-radius’ containment to a core feature via AWS sub-accounts and organizations.”
That's why Hallman believes the groundwork has already been laid for 2017.
“[There will be] a merging of the aligning trends of serverless frameworks, AWS SAM, and AWS sub-accounts in a manner that satisfies both the access needs of developers and the security requirements of DevOps.” —Dean Hallman
Hallman expects microservices and container-based cloud infrastructure to merge with serverless in 2017, rather than being positioned as a competing approach. One example of this trend is teh emergence of products like IronFunctions, from Iron.io, which have appeared recently as a "Lambda-anywhere" solution, he says.
Action items: How to move forward
Now that you have read what experts predict in DevOps this year, how should you respond? Here are a few key takeaways that should help you get your business prepared for the coming year.
- Focus on infrastructure as code in your DevOps transformations. It’s now a core component.
- Testers can still be generalists, but they need an area of specialization. They also need to know the fundamentals of programming, and have the ability to build their own applications.
- Reassess any agile processes employed by your teams, and consider going back to basics using the principles outlined in Modern Agile or Heart of Agile.
- IT operations should consider moving away from the old-style, vendor-locked PaaS services. Start exploring newer container-based and Kubernetes-based PaaS offerings to provide flexibility and portability in a hybrid environment that may contain services from multiple cloud vendors.
- Once you have researched all the reasons why you should migrate to microservices, find articles about why you shouldn’t migrate to microservices, and get a more balanced perspective to temper the hype.
- If you have only used containers in development, it's time to start experimenting with ways to use them in production.
- Development and operations management should start boning up on serverless architectures, and begin experimenting with it.
That's what the experts say. What are your predictions and retrospectives for development and operations? Let us know your opinion on these predictions and your own expectations in the comments section below.