Nutanix Gives an AI Push to End Kubernetes-Adoption Issues

BARCELONA, Spain — Nutanix executives say its AI-assisted processes and tools for developers, administrators and CIOs will represent simplified platforms and tools for user organizations. But now Nutanix is aggressively expanding its cloud native support with AI, of course.

To accomplish its historical goal, Nutanix’s efforts need to involve building momentum for open source adoption and services built on those in order to adopt and offer AI-driven solutions for operations built on Kubernetes. Not to say Nutanix has lagged in its adoption and contribution to the open source community. But for Kubernetes, at least, Nutanix’s roadmap for its new Nutanix Kubernetes Platform (NKP) heavily involves Kubernetes support.

This platform is a combination of Nutanix’s Kubernetes Engine and the main feature, the D2IQ product family, which Nutanix acquired in 2023. D2IQ was previously Mesosphere, which Tobi Knaup, Sr. Director and General Manager of Cloud Native a Nutanix, co-founded.

@nutanix’s Tony Knaup (@mesosphere co-founder) on making the leap to @kubernetesio less painful? Is Nutanix Kubernetes Platform (NKP) more than just another pain of glass for Kubernetes? pic.twitter.com/66MtfDOnOS

— BC Gain (@bcamerongain) May 21, 2024

The idea is to help users more easily get up and running and manage Kubernetes in ways they could not in the past — with automated AI input and the ability of the platform to assume related adoption and management tasks. These capabilities draw largely from the Mesosphere project and should eventually cover cluster management. As Nutanix has traditionally sought to help its users better manage the complexities of storage, data management and operations, NKP supports and automates tasks not only for cloud native but for all of its tools and services on offer. The cloud native support will, among other things, help the non-expert resources of an organization. Platform engineering teams are among the targeted users, as the platform is adapted for the hybrid developer and operations role of team members.

The management of service mesh, load balancing and networking stack, during installation and post-adoption, are Kubernetes-related tasks no non-expert would probably want to attempt. Instead, those processes “become automated” with NKP, Rajiv Ramaswami, CEO of Nutanix, told The New Stack during a one-on-one interview during .NEXT 2024 here, Nutanix’s annual user’s meeting. On a broad level, “typically, with new technology, there’s initial excitement that builds up, then subsides a bit, and then you reach a point where real applications driven by business cases and return on investment become prevalent,” Ramaswami said. “Kubernetes will continue to grow in importance and become critical to everything we do.”

Simplifies Life

Already, Nutanix’s platform solves many of the compatibility problems when working with multiple cloud vendors. For the adoption of different environments and their storage management on the cloud, the NKP does the work of the expert. Administrators will be able to operate Kubernetes across various environments without needing to hire additional experts,  Ramaswami confirmed.

“This simplifies life for administrators, allowing them to manage Kubernetes clusters using a single tool across AWS, Azure, on-prem and more,” Ramaswami said. “It also automates many workflows required to get those Kubernetes clusters up and running.”

By providing and allowing for choice, we see enterprise IT accommodating various preferences, both cloud native and not, Lee Caswell, Nutanix senior vice president of product marketing, said. “We’re doing that similarly across a number of different fronts,” Caswell said. “This approach allows us to maintain the necessary flexibility and support a diverse range of development environments.”

Kubernetes is an effective standard, but there are different development environments in which IT admins and developers work. You have people saying, “I want to use Red Hat OpenShift, while another group prefers Docker and another group might be using Rancher or working with EKS,” Caswell said. “As you offload those pieces, what we’re doing right now is basically telling our infrastructure owners, ‘hey, you’ve got this now. You can go and help your developers get access to Kubernetes containers,’” Caswell said. “We’ll still provide the degrees of freedom they require.”

In some ways, Nutanix’s NKP platform covers observability. It provides operational Insights into the performance and security of environments, cloud native and not, including on-premises An AI chatbot is used for troubleshooting. Again, this is all a lot for even a team of non-kubernetes experts to handle.

“A lot of times the skills gap is a real issue,” Knaup told The New Stack. “Together, those tools help to instantly turn you into a Kubernetes platform engineer, helping you avoid a lot of common mistakes and build out your environments according to best practices.”

More With AI

@nutanix’s Laura Jordana demoed GPT-in-a-Box for #AI: You’re able to download a model, create an endpoint, attach it to the GPUs, and then connect it to an application, on @kubernetesio in this case, with Llama 3 and an @nvidia GPU, of course. pic.twitter.com/Oh7WBnzAw5

— BC Gain (@bcamerongain) May 22, 2024

During .NEXT 2024, Nutanix showed what else its AI-powered automation can do in practice during a demo for Kubernetes infrastructure. As it is difficult to find a software provider not at least paying lip service to its AI capabilities — which are often more perfunctory than not — Nutanix demonstrated during a keynote demo during the conference.

In simple yet precise working, the demo showed “how to do a lot of things quickly,” Ramaswami said. When multiple steps are involved, LLMs become critical. Nutanix’s AI platform automatically draws LLMs from a repository that includes those Nvidia or Hugging Face provides. The user downloads the platform and attaches it to the hardware before connecting the LLM to a customer’s application.

When building an AI app, “there are a lot of moving parts,” Laura Jordana, director of technical marketing at Nutanix, explained during her demo. They include the infrastructure layer with the compute storage, the GPU, the Kubernetes platform, the LLM model and the inference server. Access control for models and data are required. “Then there is the application to develop,” Jordana said.

How Nutanix can help is the same way we’ve automated many of the complexities of managing “your storage and compute,” Jordana said. “We’re doing the same thing with GenAI… The foundation of the Nutanix cloud platform was our data services,” she said. “With GPT-in-a-Box 2.0, we’re automating all the pieces needed to create those models and different endpoints.”