Container orchestration, AIOps and CI/CD tools are familiar to enterprise DevOps pros, but expect to see them put...
to everyday use among enterprises in the new year, IT experts said.
Automation is the theme that ties these IT industry trends together, and will stem from the centralized infrastructure platforms IT ops departments put into production in 2018. These platforms will help IT practitioners at mainstream companies finally shift their focus from tactical break/fix tasks to a more strategic role as site reliability engineers (SREs).
"We're doubling down on platform and deployment automation work, and positioning the SRE team to act as an enablement function to the rest of the company," said Andy Domeier, director of technology operations at SPS Commerce, a communications network for supply chain and logistics businesses in Minneapolis. "We're looking for more consistency to what we're building and how we're operating platforms."
Container management maturity goes mainstream
The foundation for IT industry trends in 2019 lies in the fact that containers and container orchestration will finally become commonplace in enterprise IT shops. Container technology has matured to the point where it is viable for enterprise production use, mainstream enterprise vendors have standardized container management tools on the open source Kubernetes framework, and IT pros have had time to learn the technology and figure out how to use it with their corporate applications.
One of the last frontiers for container maturity in the enterprise is full-fledged container orchestration support for Microsoft applications. Docker Inc. launched a program to modernize enterprise Windows applications through its swarm mode orchestrator, and pledged to add equal support for Kubernetes on Windows once it stabilizes upstream. That stable release is now slated for the first quarter of 2019.
Meanwhile, enterprises have begun to dig in to the ways Windows workloads can benefit from containers, a technology originally invented for Linux.
Richard Fongsenior software engineering manager, Mitchell International
"Windows has been lagging Linux in containers, but now it may be able to leapfrog ahead," said Richard Fong, senior software engineering manager at Mitchell International, an auto insurance software company in San Diego. "We're ironing out how to do DevOps automation for Windows, where container images are still bigger than on lightweight Linux operating systems."
Container maturity also means that container management products, and the skills to use them effectively, will both be in high demand in 2019. The market for container management tools will swell to $2.1 billion in 2019, and reach $4.3 billion by 2022, estimates 451 Research in its "Market Monitor" study on application containers issued in December 2018. And interest in Kubernetes among job searchers jumped 173% in 2018 compared to 2017, according to job listings site Indeed's annual "Occupation Spotlight" report.
AIOps gets real
As DevOps shops break down enterprise applications into microservices and deploy them onto container platforms, the number of moving parts within them increases exponentially, as does the complexity of incident response. Artificial intelligence for IT operations, or AIOps, has been a trendy topic among IT monitoring vendors since late 2017, but those products have struggled to live up to their claims, enterprise IT buyers said.
In part, this has to do with product maturity, but also improvements in data management in the repositories of information those tools analyze.
"The data we have now on IT systems is cleaner than ever, because it's generated through automated systems at the speed of the digital world, rather than humans," said a senior architect at a large insurance company on the East Coast, who spoke on condition of anonymity because he is not authorized to represent his employer in the press. "Many tool vendors also advertised [AIOps] capabilities to gain attention before they were really mature, but that's shifted to tangible things we can try now versus slideware."
The senior architect declined to name the vendors his company will consider, but other DevOps pros have their eye on machine learning features in application performance monitoring tools from vendors such as New Relic and Instana.
"We've been building app profiles to work with anomaly detection systems, making sure they're emitting the right messages," said Brad Linder, DevOps and big data evangelist at Dish Technologies, the engineering arm of Dish Network in Englewood, Colo. The firm will review AIOps tools from Instana and its competitors in 2019, he said.
Some SREs argue that platforms should be 'boring' -- made of simple components and deployed in repeatable patterns to reduce the need for machine learning analysis. But Linder said he still sees AIOps tools as useful, regardless of how highly automated the infrastructure becomes.
"The platforms are boring, but there are still glitches in the matrix to watch out for," he said. "Also, AI can be used to detect good things -- application services that are more popular among users than we anticipated, for example."
AI will also become a common feature of database operations.
"Oracle has already begun offering cloud databases that it calls autonomous, and I expect that in the coming year that you'll see the other big database [vendors] start adding [machine learning] to make databases more self-driven," said Tony Baer, an analyst at Ovum. "The trend will start in the cloud where the database vendor has control of the infrastructure, but I know of at least one database player that will start offering some self-running capabilities on prem."
DevOps teams test fresh takes on CI/CD
As container management and AI improve IT infrastructure automation, application deployment tools will evolve to capitalize on those advances. Thus, enterprise DevOps teams will explore new approaches to CI/CD in 2019 that surpass traditional systems such as Jenkins.
Some mainstream companies have already switched in 2018 from Jenkins to Netflix's Spinnaker platform, primarily because it supports automated canary application deployment patterns and rollbacks. Other emerging* CI/CD tools that build on container infrastructure have made it onto the enterprise DevOps radar, such as Drone.io, Harness, and Intuit's Argo CD.
"Spinnaker works, but it looks like a big hammer with lots of options that's a bit too heavy for my liking," Linder said. Instead, he said he's interested in Harness for its data analytics features.
SPS's Domeier, meanwhile, said his company has moved from Jenkins to Drone.io for CI/CD.
"We had more and more problems with our old Jenkins patterns and pipelines," he said. "Drone configuration is done as code and it's more container-native [than Jenkins] -- we would be unlikely to use it if we were trying to ship large JVMs or WAR files that weren't containerized."
SPS engineers favor Drone's smooth integration with GitHub code repositories, Domeier added. The Jenkins community also offers Jenkins X, which improves its integration with Kubernetes.
* Information changed after publication; only Drone.io is open-source.