Containers are growing in popularity because they fit well with modern development techniques, such as DevOps, but many businesses find they need new management tools. The market for container management tools is in a fledgling stage, with many options, but no clear leader.
"A lot of different pieces are coming together and driving new opportunities, as well as creating new challenges [with container management]," said Stephen Elliot, program vice president at IDC.
Much like virtualization, containers move application development one layer away from the computing infrastructure, so developers spend more time enhancing applications and less time configuring system resources. This approach fits with today's continuous development mantra, which stresses speed.
Consequently, containers are becoming quite popular. In fact, the application containers market will create $4.3 billion in revenue in 2022, according to 451 Research's November 2018 Market Monitor study on application containers.
New management requirements
When companies deploy containers, many find they require new container management tools to use them at scale.
"Monitoring container platforms means managing numerous moving parts, from a security, performance, compliance and availability perspective," said Torsten Volk, managing research director at Enterprise Management Associates (EMA) in Boulder, Colo.
Existing virtualization management tools often fall short of organizations' needs, because multiple containers can share the same host OS. Another management challenge stems from the fact that containers are often used for brief periods -- in some cases, only a few seconds. Tracking container performance and the effect on system resources is a challenge.
Vendors have made progress in addressing some user needs.
"Three or four years ago, companies had to build out their own erector sets to manage containers, and few organizations are capable of doing that," said Dennis Smith, research vice president at Gartner.
In 2019, a number of fast-moving, open source projects provide building blocks, so organizations can manage these elements in a simpler manner. Kubernetes has emerged as the leader for orchestration. Leaders in other areas, such as performance monitoring, are taking shape, but aren't quite as clear.
"It's a bit of a free-for-all among a number of different open source projects," Smith said.
As a result, vendors use different approaches to deliver container management tools. VMware and IBM aim to provide a single operating model for both VMs and containers; Docker and Rancher are among the popular Kubernetes-based management platforms; and other vendors, such as Mesosphere, attempt to turn data centers and clouds into scale-out resource pools, EMA's Volk said. In addition, a bevy of other suppliers, such as AppDynamics, Dynatrace, New Relic, Platform9, Sensu, SignalFx and Sysdig, offer various niche management platforms.
What organizations need to move forward
Systems management consists of a variety of functions, and the various container management tools are strong in some areas and weak in others. They provide basic scheduling functionality, so a business understands when its applications will execute. Increasingly, they also help tune how well those system resources perform. When problems such as downtime arise, many available tools often automatically roll back processing to simplify recovery.
Stephen Elliotprogram vice president at IDC
Those features are noteworthy, but plenty of desired functionality is missing.
"Clearly, the management features available with containers are not nearly as robust as the tools found with virtual machines," Gartner's Smith said.
When a performance monitoring problem arises -- say, intermittent slow response times -- troubleshooting can be a challenge.
"Containers present IT with a double-edged sword," IDC's Elliot said. "They are easy to create, but also easy to destroy. Large companies may generate billions of containers each week, which becomes difficult to track."
To date, IT administrators use containers and management tools for relatively small deployments, and there's still a question of whether these tools will be able to scale up to support the large deployments that sophisticated applications require.
Another issue is integrating container management tools with existing hardware and software. Although containers work across many different environments, there are plenty of limitations when connecting them to legacy systems. For example, Windows and Linux containers aren't interoperable.
So, how are companies dealing with these management limitations?
"The irony is that many enterprises -- maybe most -- now run their containers on top of virtual machines," Gartner's Smith said.
Over the last decade, virtualization management tools have matured into robust, comprehensive systems management tools. Container management tools will gradually improve, but for now, companies that run containers at scale will need to recognize that the market is still evolving.