VMware Embraces Kubernetes APIs in VMware vSphere 8
At the VMworld Explore conference this week, VMware announced that the latest version of the VMware vSphere platform now supports the Kubernetes application programming interface (API). That support enables DevOps teams to streamline the management of virtual machines and Kubernetes clusters.
Included as part of a VMware Tanzu Kubernetes Grid 2.0 update that is part of vSphere 8, the goal is to reduce the number of APIs that DevOps teams need to employ when deploying Kubernetes clusters within a vSphere environment.
As part of that effort, VMware has added a Cloud Consumption Interface service that provides access to both graphic interfaces and APIs to provision virtual machines and Kubernetes clusters, including a simplified cluster creation API, dubbed Cluster Class, and a set of tools based on the open source Carvel project.
The company is also now making it possible to create an availability zone spanning three clusters running VMware vSphere 8 along with adding support for data processing units (DPUs), also known as smart network interface cards (SmartNICs), to offload the processing of network and security tasks using a VMware vSphere Distributed Services Engine.
VMware CEO Raghu Raghuram told conference attendees that DPUs, alongside CPUs and graphical processor units (GPUs), are the foundation upon which the next decade of modern computing will be built.
Finally, vSphere 8 adds support for up to 32 NVIDIA GPU devices in passthrough mode to increase the performance of artificial intelligence (AI) models.
IT teams that deploy Kubernetes clusters on virtual machines routinely make use of a mix of APIs and graphical tools to manage a complex IT environment. That approach enables DevOps teams to work alongside IT administrators that don’t always have the programming skills required to invoke an API. In general, VMware is making a case for an approach to building and deploying cloud-native applications that does not always require organizations to hire full-stack developers and site reliability engineers (SREs) to manage the underlying platform. The goal is to make Kubernetes environments more accessible to the average IT administrator at a time when DevOps professionals are hard to find and retain.
It’s too early to say how heavily IT organizations will rely on dedicated DevOps teams to manage Kubernetes environments versus employing a “team sport” approach that enables engineers and administrators to work more collaboratively. The relationship between DevOps teams and centralized IT organizations continues to steadily evolve. Rather than relying on custom scripts that don’t usually scale and are generally poorly documented, APIs defined and maintained by the open source community are increasingly employed to provision infrastructure at scale. That approach makes it simpler for IT organizations to employ multiple distributions of Kubernetes clusters as they see fit.
It is clear that VMware is embracing de facto open source API standards rather than employing proprietary APIs that would lock customers into a specific management framework. In effect, the ‘not invented here’ syndrome that resulted in VMware’s reluctance to embrace open source software appears to be coming to an end.
