Diamanti has made available a technology preview of a control plane for hybrid deployments of Kubernetes clusters spanning public and private clouds.
Company CEO Tom Barton says Diamanti Spektra will extend the reach of Diamanti beyond the bare-metal appliances optimized for on-premises deployments of Kubernetes the company delivers today.
Diamanti originally developed a control plane for managing fleets of bare-metal appliances running Kubernetes. Diamanti Spektra builds on those capabilities to make it possible to move containerized applications to public clouds running any distribution of Kubernetes, Barton says, noting as such, one of its primary use cases will be disaster recovery.
However, he adds, one of the primary reasons any IT organization invests in Kubernetes is to eventually drive hybrid cloud computing strategies. Today most IT organizations typically employ multiple clouds that are managed in isolation. Kubernetes presents an opportunity to employ a common layer of abstraction across public and private clouds running in an on-premises IT environment. To achieve that goal, however, a common control plane must be employed across multiple distributions of Kubernetes.
Longer term, Diamanti expects to be able to extend that control plane to other X86-based platforms running in on-premises IT environments as well, Barton says.
Diamanti is one of many vendors trying to leverage Kubernetes to drive hybrid cloud computing scenarios. Diamanti Spektra, however, will be unique in that it enables internal IT teams to retain ownership of that control plane versus ceding control of their Kubernetes environments to a cloud service provider, he says. Otherwise, IT organizations will run the risk of finding themselves locked into a specific control plane service.
Less clear is how many of those instances of Kubernetes will be running on virtual machines versus bare-metal servers. Most cloud service providers still rely on virtual machines to provide isolation between instances of Kubernetes. However, in on-premises environments, many IT organizations are moving away from virtual machines to improve the performance of applications. At the same time, Intel is working on an approach to embedding hypervisors in processors that would make it possible to dynamically invoke different classes of hypervisors for more legacy applications and emerging containerized applications. That approach would make it possible to employ a lightweight hypervisor that would primarily serve to isolate instances of Kubernetes without adding much processing overhead.
However the battle for control over Kubernetes deployments plays out, the distinction between clouds running on public platforms and on-premises environments in the age of Kubernetes is about to dissipate. If every platform is running an instance of Kubernetes, managing one platform versus another shouldn’t be fundamentally different. In the meantime, Diamanti and a host of other IT vendors are now racing to become the control plane that manages all those cloud platforms regardless of where they might be located.