Alongside the formal release of version 1.0 of the open source Kubernetes orchestration framework for containers, CoreOS today at the OSCON 2015 conference unveiled Tectonic Preview, a version of its Linux distribution through which the startup will provide 24/7 support for IT organizations using Kubernetes to manage Docker containers on the CoreOS platform.
CoreOS CEO Alexi Polvi said at this juncture open-source Kubernetes, originally developed by Google, will emerge as the de facto standard for managing containers across hybrid cloud computing environments.
“Google has the largest deployment of containers running in a production environment,” said Polvi. “I think most organizations will want to use the orchestration framework they developed. We’re betting the company on it.”
Polvi said the most significant thing about the arrival of Kubernetes 1.0 is that hybrid cloud computing becomes practical for IT operation teams, because there is now a framework for managing Docker containers on multiple platforms. That capability, he added, is also likely to prove critical to Google’s cloud business because applications developed on the Google Compute Engine platform can be moved on-premise or vice versa, moved into the Google cloud.
Priced starting at $1,500 a month, IT organizations can opt to deploy CoreOS on-premise, in a hosted environment or in the cloud, Polvi said. CoreOS will also be working with Google and Intel to provide additional Kubernetes training.
While it’s unclear how often Docker containers will be deployed on bare metal servers versus VMs, Polvi said it may not matter. Docker containers are about creating a single pool of resources that can be made up of both physical and virtual machines.
Conversely, VMs are about slicing a larger physical server into many smaller machines. As a result, Polvi contends that Docker containers are really a superset of VMs.
What is almost for certain is that there will soon be many more Docker containers than VMs. As Docker sprawl continues to grow across multiple data centers, IT organizations are going to need orchestration software to manage it all, regardless of whether those containers are running on a bare-metal server, a VM, or some private or public Platform-as-a-Service environment.
In fact, Polvi said, the primary challenge may not necessarily be the volume of containers that needs to be managed but rather the velocity at which those containers will come and go inside the data center.