For IT teams inside large organizations used to managing any number of operating environments, Kubernetes is a breath of fresh, standardizing air. Forget its origins, forget any excitement over containers or microservices, and forget the sprawling ecosystem of related projects. What has some folks charged with managing Kubernetes deployments really excited is the prospect of managing all application infrastructure essentially the same way.
Many apps, one abstraction
According to Juergen Sussner, a cloud platform architect at German IT service provider DATEV, Kubernetes has the potential to be a game-changer for how teams like his do their jobs.
“Before we had the standardization layer, we had a lot of things to consider when putting new software in the data center," he says, "like about network topology, sizing of VMs, how to place the VM into the network, firewalling, and all this stuff. And nowadays you can say, ‘Does your product run on Kubernetes, do you have a Helm chart for the plan deployment?’ for example. If yes, we're fine. If no, maybe we choose another one.”
The result is that commercial-off-the-shelf (COTS) software, middleware, and homegrown software all share a management experience at the infrastructure level. Sussner’s team provides a stable platform for delivering Kubernetes clusters as a service—including by implementing security tooling and other required functionality, as well as robust documentation—and other teams build what they want on top of the Kubernetes cluster without having to involve operations at every step along the way.
Of course, this is especially beneficial for application developers who want to be able to move fast without worrying about breaking anything.
“Our objective as a cloud platform team for DATEV,” says Sussner, “is basically to get out of developers’ way. We want to get the developers as productive as possible by standardizing infrastructure, by creating environments where developers can push applications to production without the need of infrastructure guys.”
And as far as they’re concerned, the only reason they can do that is because both the environments and components are standardized.
“VMware evolving into an open source company, or towards an open source strategy, is really promising for me. You have that huge open source community and that is the real power of all these Tanzu products.”
—Juergen Sussner, cloud platform architect at DATEV
Multiple teams, multiple clusters, one management platform
Robert Stamps, a senior cloud architect at TD Ameritrade, is also keen on using standardization—as well as automation—as a way to help business units focus on applications rather than on infrastructure. However, despite having multiple Kubernetes clusters in place, his team isn’t there quite yet.
“I can only run the Kubernetes commands across all the clusters so many times,” he says, jokingly adding, “I only have so many keystrokes in life, and they’re running low.”
One of the issues is the regulated nature of the financial services industry in which TD Ameritrade operates. While the DevOps teams within mature business units can more or less keep their own clusters updated and compliant (perhaps with a little nudge to let them know, for example, that their operating systems are out of date), other teams would rather someone else do that work for them. And then there are tasks such as monitoring cluster access, rapidly patching CVEs, and generally making sure the company meets the security expectations of external auditors.
“I can only run the Kubernetes commands across all the clusters so many times. I only have so many keystrokes in life, and they’re running low."
—Robert Stamps, senior cloud architect at TD Ameritrade
A standard, automated process for monitoring clusters and carrying out these functions would make life much easier. Those are some of the benefits TD Ameritrade sees from running other applications on Tanzu Application Service (TAS), and Stamps says he’d love to see that type of automation come to the company’s Kubernetes environments. (That, in fact, is the world VMware is targeting with Tanzu Mission Control: a centralized Kubernetes management platform that supports multiple environments and distributions by baking Kubernetes into the core of TAS.)
“Automation is really used to control the massive expansion of staff that can happen when you start getting involved with Kubernetes,” says Stamps. “So we’re trying very consciously to understand that and prioritize where we spend our time automating things.”
One API to rule them all
DATEV’s Sussner also sees value in Kubernetes, which is now baked into vSphere, becoming the standard for managing vSphere environments. Programmatic infrastructure based on modern APIs is the de facto norm for public cloud-based resources, and that’s exactly what Kubernetes now brings to vSphere environments, whether they’re based on-premises (like DATEV’s are) or in the cloud.
“What I find interesting is the Kubernetes interface to vSphere, the possibility to use a standardized Kubernetes control language to interface with the classic vSphere topology to create a VM with a Kubernetes command,” says Sussner.
“[W]ith the Kubernetes capabilities, I can include it in any pipeline that is Kubernetes-enabled and I can use that pipeline to create on-demand resources I need for a test—for integration tests, for example—and destroy the resources afterwards, using the same commands I would use to operate my Kubernetes clusters.”
“Developers are pushing this new technology to their operations or infrastructure department, and the operations teams are coming to me for answers.”
—Christoph Villnow, senior IT architect at Unique Projects
Education and community: Soon, we’ll all know k8s
If Kubernetes has one primary weakness preventing it from reaching relative ubiquity and becoming a de facto standard, it might be that the learning curve can be steep. Christoph Villnow, of German IT consultancy Unique Projects, says his firm and its clients are just getting on board with Kubernetes, and everyone could use some education on how it works and how it changes their jobs. “Developers are pushing this new technology to their operations or infrastructure department,” he says, “and the operations teams are coming to me for answers.”
What’s most necessary, he surmises, “is to increase the knowledge of how to move from the old on-premises server infrastructure into all the new things Kubernetes brings with it.”
However, learning Kubernetes is getting easier thanks to its fast-growing community. “I think the community around these technologies like Kubernetes is the most important thing, because there is this big source of knowledge in the community,” says DATEV’s Sussner. “And if you have a question, reach out to the community and someone else will help you. You will get your answer.”
The number of vendor-led initiatives that marry the open source ethos with a product development point of view—the VMware Tanzu line of products, for example—will also help cloud native newcomers get up to speed and into production faster.
“VMware evolving into an open source company, or towards an open source strategy, is really promising for me,” says Sussner. “You have that huge open source community and that is the real power of all these Tanzu products.”
