Best of 2020: OpenStack’s Complicated Kubernetes Relationship
As we close out 2020, we at Container Journal wanted to highlight the five most popular articles of the year. Following is the second in our series of the Best of 2020.
2020 may be the year the OpenStack community comes to terms with Kubernetes
As the open source community heads into 2020, loyalties between OpenStack and Kubernetes are likely to become increasingly divided. Contributors to open source projects are trying to determine where to prioritize their efforts, while IT organizations are wondering to what degree they will need a framework such as OpenStack to deploy Kubernetes.
Most Kubernetes deployments thus far have been on top of open source virtual machines or commercial platforms from VMware. Most of those decisions have been driven by the need to isolate Kubernetes environments sharing the same infrastructure. In addition, many IT organizations lacked the tools or expertise required to manage Kubernetes natively, so it became easier to simply extend existing tools to manage Kubernetes as an extension of a virtual machine-based platform.
The debate now is to what degree that approach will continue as organizations become first more familiar with native Kubernetes toolsets and alternative approaches to isolating workloads using lighter-weight virtual machines emerge.
Lighter-weight alternatives to OpenStack and VMware for deploying Kubernetes clusters already exist, notes Rob Hirschfeld, CEO of RackN, a provider of an infrastructure automation platform based on open source Digital Rebar software.
At the same time, managed service providers such as Mirantis have begun rolling out highly distributed services based on Kubernetes that make no use of OpenStack at all.
Jonathan Bryce, executive director for the OpenStack Foundation (OSF), says he expects OpenStack and Kubernetes to continue to evolve in a complementary fashion. Organizations that have invested in OpenStack are not likely to abandon their investment, he says.
In addition, the OpenStack community has been developing Airship, a collection of open source tools for automating the provisioning of cloud infrastructure using a declarative framework, while also driving the development of Kata containers, an open source container runtime designed to run on a lightweight instance of a virtual machine as an alternative to either Kernel-based Virtual Machines (KVM) or VMware vSphere platforms. In fact, Bryce says that as far as platforms such as Kubernetes are concerned, the OpenStack community as a whole does not tend to be “super religious.”
In many ways, the OpenStack community is confronting many of the same challenges that face VMware. As Kubernetes continues its momentum, IT organizations are trying to determine whether they will be able to rely solely on a cloud-native Kubernetes stack versus continuing to layer Kubernetes on top of their existing stack of legacy IT infrastructure. It may turn out there will be instances of Kubernetes running on top of legacy virtual machines for many years to come. At the same time, however, IT organizations may also decide to deploy a growing number of greenfield cloud-native native applications on Kubernetes clusters running on either light-weight virtual machines or bare-metal servers.
Regardless of the path chosen, the various ecosystems on which Kubernetes clusters will be deployed will be nothing if not diverse.