With Kubernetes having officially celebrated its 6th birthday as an open source project, it’s apparent just how profound an impact the platform is about to have on enterprise IT in terms of both how IT infrastructure is managed and applications are deployed.
As a percentage of the IT platforms deployed in production environments by enterprise IT teams, Kubernetes use is comparatively light, even after six years of contributions to the open source platform. However, it’s arguable Kubernetes really became enterprise-ready just in the last two years. As such, Kubernetes is just now coming into its own.
From an infrastructure perspective, the biggest impact Kubernetes has had is accelerating a transition to hyperconverged infrastructure (HCI) platforms. Kubernetes assumes an HCI platform that unifies compute, storage and, to some degree, networking has already been put in place. Developers looking to deploy microservices-based applications based on containers that are orchestrated by Kubernetes are forcing IT operations teams to accelerate a transition to HCI platforms, many of which began earlier this decade.
HCI platforms are most widely employed by cloud service providers, which makes it easier for DevOps teams to automate provisioning Kubernetes clusters. Many internal enterprise IT teams are now following suit by adopting frameworks to automate the management of Kubernetes clusters running on HCI platforms deployed in on-premises IT environments.
The bulk of Kubernetes clusters deployed today are in public clouds. However, as Kubernetes becomes easier to manage using automated frameworks, the number of Kubernetes clusters running in on-premises IT environments will increase, says Haseeb Budhani, CEO of Rafay Systems, a provider of IT tools for managing Kubernetes clusters. In fact, given the inherent complexity of Kubernetes, the platform is driving increased demand for IT automation frameworks.
“Automation has never been more important,” he says.
Budhani also notes there always will be applications that perform better in an on-premises IT environment than in the cloud. Plus, security concerns and compliance requirements prevent many workloads from being deployed on a public cloud.
Additionally, many more IT organizations that need to deploy applications in a local data center are under pressure to reduce costs in the wake of the economic downturn brought on by the COVID-19 pandemic, says Jenny Fong, vice president of marketing for Diamanti, a provider of hyperconverged appliances optimized to run Kubernetes.
“A lot of organizations are reassessing IT budgets,” she says.
Organizations that opt to deploy workloads on bare-metal servers running Kubernetes can eliminate the need to license commercial virtual machine software from VMware, Fong notes.
As Kubernetes adoption in on-premises IT environments increases, it will become the underlying infrastructure that enables IT organizations to fully embrace hybrid cloud computing, says Tina Nolte, vice president of product for Spectro Cloud, provider of an IT framework for managing Kubernetes clusters.
The challenge many IT teams soon will face is finding a way to manage fleets of Kubernetes at scale running on multiple platforms that are likely to be running different versions of Kubernetes and different distributions, adds Nolte.
“IT is going to struggle with updates,” she says.
As Kubernetes is installed in more organizations, the rate at which microservices-based applications based on containers will be deployed on those clusters will also increase, especially as the application development environments needed to build those applications are more widely adopted.
Red Hat has been at the forefront, with its OpenShift platform-as-a-service (PaaS) environment that runs natively on Kubernetes. That environment is designed to enable developers to work with Kubernetes directly via kubectl command-line interface (CLI) and application programming interfaces (APIs), other CLIs or via integrated development environment (IDEs).
The goal is to meet developers anywhere they are along the spectrum by providing access to optional interfaces, says Clayton Coleman, chief architect for OpenShift at Red Hat.
“How people build applications is so varied,” he says.
Coleman adds that while Kubernetes is wildly more successful than initially thought it would be, the original vision for the platform—a platform on which other platforms are constructed—has not yet been attained. Coleman says there is a clear need for higher levels of abstraction to automate the management of such platforms.
Competition among providers of application development and deployment platforms that provide that level of abstraction natively on Kubernetes clusters is now heating up. The Cloud Foundry Foundation (CFF), for example, has outlined a plan to accelerate transitioning its namesake open source PaaS to Kubernetes, and distributions of which are expected to be available in 2021.
Those distributions will provide IT teams with an opinionated development environment that is already widely employed in many large enterprise IT organizations, says Chip Childers, executive director of the CFF. In fact, he notes, one of the things holding back the adoption of Kubernetes in the enterprise has been a general lack of development tools for the platform.
“There’s been too much focus on infrastructure rather than the developer experience,” Childers says.
In effect, Kubernetes transforms everything above and below it.
There’s no doubt the number of microservices-based applications running on Kubernetes clusters is about the explode. Regardless of where those applications are deployed, however, the next big challenge awaiting IT teams will be finding a way to manage a raft of next-generation cloud-native applications that are just as likely to be deployed in on-premises environments as they are the cloud.