2016 Dev Predictions, Part 3: Mainstream Microservices, Reactive Streams and Containers-as-a-Service
I know, it's February, but I reached out to a lot of smart people last month for their thoughts on what lies ahead for enterprise developers in 2016 -- more than I could squeeze into parts 1 and 2. In addition to industry analysts, I connected with some of the thoughtful execs I spoke with last year.
Martijn Verburg, for example, sent me an intriguing list of predictions. Verburg is the CEO of jClarity, which focuses on automating optimization for Java and JVM-related technologies, and he serves as co-leader of the London Java Users Group. Last year he predicted that developers would begin to feel a new pressure to fully understand DevOps and virtualization, "or be left scratching their heads when their apps fail miserably in their private or public cloud deployments." This year he added containerization to the trends that will stress developers in 2016.
"Docker has now moved beyond the mass adoption curve and is more or less here to stay," he said, "although many SA's won't touch it for production. Interestingly some of the DevOps tooling (Chef, Puppet, Ansible) will become less important as developers cobble together containers running services instead. For example, we used to Chef up a Host with Java, MongoDB, and a host of other software. Now we just deploy a set of Docker containers that all contain one major service each (MongoDB, or Java or Foobar)."
"The wise developer who really starts to utilize Docker in test will be the productive developer," he added. "No more relying on QA infrastructure, just bring up and down containers as you need! Extremely wise developers will not rely on the results of performance testing on container based architectures, unless they mimic production."
He also believes that microarchitectures "will continue to be much hyped, seriously abused, and difficult to co-ordinate and maintain." But we should also expect to see some good tooling emerge by the end of the year.
Jonas Bonér also sent me a list of prognostications. Bonér is the CTO and co-founder of Typesafe, the company behind Scala, the general purpose, multi-paradigm language that runs on the Java Virtual Machine (JVM), and Akka, the open-source run-time toolkit for concurrency and scalability on the JVM. Last year he said that he expected container-based infrastructures to continue to make life easier for the developers who adopt them. He pointed to several technologies driving this trend, including Docker, Apache Mesos, Google Kubernetes, and CoreOS. He also said that the game-changing impact of Java 8 would continue throughout 2015 as adoption spread -- which would have the interesting side effect of allowing more functional style programming.
In the coming year, Bonér expects microservices to "graduate" from an early adopter tool to "the first wave" of real mainstream adoption. The reason: The technical constraints that held microservices back -- things like single machines running single core processors, slow networks, expensive disks, expensive RAM, and organizations structured as monoliths -- are gone.
"Networks are fast, disks are cheap (and a lot faster), RAM is cheap, multi-core processors are cheap, and cloud architectures are revolutionizing how we design and deploy systems," he said in an e-mail. "In 2016 we have a much more refined foundation for isolation of services, using virtualization, Linux Containers (LXC), Docker, Unikernels, and Reactive runtimes like Akka. This has made it possible to treat isolation as a first class concern -- a necessity for resilience, scalability, continuous delivery and operations efficiency -- and has paved the way for the rising interest in microservices-based architectures, allowing you to slice up the monolith and develop, deploy, run, scale and manage the services independently of each other."
"The need for building Reactive applications is driving people towards microservices," he added, "and Reactive power users in particular already claim a very high incidence of building microservices." To support that claim, he points to a recent survey, the results of which are post on the company's Web site. (It's well worth reading.)
Bonér also expects a growing number of "Fast Data" tools and libraries to embrace Reactive Streams, an initiative to provide a standard for asynchronous stream processing with non-blocking back pressure. He pointed to such Reactive Streams-compatible products as Akka Streams, upcoming Java 9 Flow and Spark Streaming, Gearpump, Jetty, Vert.x and Cassandra.
Verburg also sees Reactive/Dataflow style applications becoming increasingly popular in the coming year, but also increasingly misunderstood, because of a lack of visual tooling support. He also recommended the Vert.x tool for Java JVM developers.
I also received an interesting list of predictions from Docker, and I had a chance to talk about them with Scott Johnston, the company's SVP of Product Management. His company sees containers becoming "the prevalent mechanism for application development and deployment" in the coming year, thanks to advancements in 2015 around container security, manageability, storage, networking, which are driving rapid growth. He pointed to an O'Reilly report ("The State of Containers and the Docker Ecosystem 2015," which shows that 40 percent of organizations using Docker have it in production currently, and that those numbers are expected to rise sharply in the coming year. (Another report you should read.) He also pointed out that downloads of Docker images have risen from 67 million in December 2014 to 1.2 billion last year.
Perhaps Johnston's most interesting prediction: the rise of Container-as-a-Service (CaaS) architectures, which will facilitate Ops-originated application delivery.
"Think of Containers-as-a-Service as a platform for development teams to get the agility they need to build, ship, and deploy containers, while giving Ops the control they need to adhere to governing and regulatory standards and uptime SLAs in the datacenter," Johnston told me. "It's the notion of Ops standing up infrastructure that is container-ready and aware, while providing a self-service interface for the development teams to use to pick up the containers they've made and deploy them and manage them in production. We're trying to balance the agility needs of the Dev organization with the control needs of the Ops org."
"CaaS will succeed without requiring organizational changes as seen with the rise of DevOps," the company stated, "eliminating the need to retool and re-skill by refocusing on what Ops can do for Dev through integration of core and container technologies, thereby creating a more circular pattern of collaboration."
"At the end of the day, whether you're in Dev or Ops, the business of IT is shipping apps that deliver value to the business," Johnston said. "So it helps for devs to carry pagers, to make sure the apps they're shipping maintain the quality they need, and thereby get a little taste of the Ops world. And it makes sense for Ops to push toward the Dev side, to get their hands a little dirty with code so they understand how the code is working, so they can make sure they create environments that will stand it up."
Johnston said we should expect to see the balance between Dev and Ops—between agility and control—improve in the coming year as container-based services become more "Ops-led" and less of a Dev-only model. Dev and Ops will share the development lifecycle, his company predicts, with Ops setting up development environments in which everything, from security to management, is baked into the platform. And containers will move beyond Dev and Test to become a production mainstay, thanks to the integration of enabling technologies from the container ecosystem, as well as accelerated innovation from de-facto container leaders.
"Apps are increasingly where the value is," Johnston said, "and lower-in-the-stack layers are increasingly becoming commodities. In some ways that's just a continuation of that trend we've seen with hardware and operating systems."
More 2016 enterprise development predictions:
Posted by John K. Waters on February 16, 2016 at 1:41 PM