When Containers Get Complicated, Shift to Kubernetes as a Service

Instead of orchestrating container technologies in a private data center, businesses can use Kubernetes as a cloud service, which provides an up-to-date IT environment across private, hybrid and multicloud systems.

By Gene Knauer

By Gene Knauer February 03, 2021

Packing items into a container and shipping them overseas is nothing new, but it’s become all the rage in the digital world of business applications. Building a business application in a container allows companies to scale, update or relocate between private and multiple public clouds with ease — so easy that containers multiply exponentially. That’s where container orchestration, or Kubernetes, comes into play. Still this technology can be overkill or quickly grow overly complicated for many businesses, and this is giving rise to Kubernetes as a service.

Kubernetes is like a band orchestrator that determines which instruments (in this case, containers) play together, for how long and when to go silent. The technology automates the setup, monitoring and management of containers that encase business applications. Originally developed at Google and named after the Greek word κυβερνήτης, (helmsman or pilot), Kubernetes is an open-source orchestration system that makes it easier to organize and schedule applications across private, hybrid and multicloud IT operations.

According to research by Datadog, Kubernetes has become the de-facto standard for container orchestration. Its adoption has doubled between 2017 and 2020 and 451 Research has found that 57% of enterprises are actively adopting Kubernetes.

“Excitement around Kubernetes has proven to be its biggest undoing,” believes Satyam Vaghani, vice president and general manager of IoT and artificial intelligence at Nutanix. 

Related

The Disruptive Force of Cloud Native

That’s because, for many, Kubernetes has become more complex than an Agatha Christie mystery novel. New versions keep coming every quarter and with them is a constellation of infrastructure options, security challenges and IT staffing pressures. Some point out that Kubernetes acts like it’s the only platform that can run applications in a system. 

For those caught up by the excitement but not interested in all the fuss, Vaghani points to Kubernetes as a platform-as-a-service (PaaS), especially for businesses that manage applications across private and multicloud environments.

Why Kubernetes is so Popular ─ and Complex

Kubernetes has been called the "Linux of the cloud" because, like an operating system, it handles many operational tasks across clouds, like assigning containers to machines, booting the containers, and handling upgrades, rollbacks and failures. It also creates cluster resources like service discovery, inter-virtual machine networking, and cluster ingress and egress.

But Joe Beda, one of Kubernetes creators, admitted recently in InfoWorld that Kubernetes is a complex system and entails a steep learning curve. Along with the complexity of using Kubernetes effectively is the diversity of implementations, with 21 versions of Kubernetes in use and new minor releases every quarter. Kubernetes versions that are certified by the Cloud Native Computing Foundation (CNCF) ensure interoperability and no vendor lock-in but many software and cloud platform vendors provide their own branded versions of Kubernetes, further increasing the differences between one deployment and the next and developer skill sets from company to company.

Why is Kubernetes so challenging to Implement?

  • Security challenges are multiplied as traffic (and the attack surface) increases between containers in the same pod, deployed together on the same host. The distributed, dynamic characteristics of Kubernetes don’t work well with legacy security solutions.
  • Complexity, different versions, and its fast evolution make Kubernetes challenging for in-house personnel to stay up-to-date on. It is evolving quickly, along with its growing ecosystem of related technologies and is a distributed system that requires new compute, storage, and networking that must adapt to how Kubernetes operates.
  • Multicloud usage is difficult to optimize, requiring different tools for visibility and control of deployments on different clouds.

“People are spending an insane amount of time in Kubernetes experiments,” said Vaghani. “Armies of developers are becoming enamored with different and newer versions. Older versions wind up unsupported after three months.”

Related

Rise of Computing to Feed the Needs of Machines

He said IoT and edge use cases are where some Kubernetes deployments are running into trouble, with developers trying to fit newer, code-heavy versions of Kubernetes on tiny devices that require miniaturization.

“If you want to deploy airport kiosks for self-check-in, for example, you shouldn’t have to spend your time worrying about whether or not this constantly evolving and changing technology will run well at the edge on tens of thousands of kiosks,” Vaghani said. 

“You don’t want your software engineers spending 90% of their time to upgrade, test and secure Kubernetes regularly. Then they can get stuck on different networking options, load balancing options, storage options and management options. You want them focusing on designing the best apps.”

Why PaaS for Kubernetes

In the past, vendors provided the database, storage system or hypervisor, according to Vaghani. But in the era of virtualization, containerized apps and microservices, the software comprises the operating layer.

“It provides the false hope to many developers that they can assemble the operating layer themselves,” he said. “But while these containerized apps and microservices technologies are now more accessible and open source, the complexity hasn’t gone away.”

That’s the appeal of Kubernetes as a cloud service. 

“First, you can enjoy the traditional benefits of Kubernetes: faster time-to-production, non-disruptive live upgrades, higher server efficiencies, elastic scalability, and application portability across different environments,” Vaghani said. 

“Second, a Kubernetes PaaS eliminates all of the work inhouse developers have to do to deploy, manage, and maintain Kubernetes across the data center and multiple clouds.” 

A single pane of glass provides visibility into and control of all Kubernetes deployments.

For these reasons ─ plus the time, cost, and uncertainty of DIY ─ Vaghani said that businesses are seeing the value in Kubernetes as PaaS.

Gene Knauer is a contributing writer who specializes in IT and business topics. He is also the author of  Herding Goldfish: The Professional Content Marketing Writer in an Age of Digital Media and Short Attention Spans.

© 2021 Nutanix, Inc. All rights reserved. For additional legal information, please go here.