Kubernetes, Docker ed il futuro dei container

Un'intervista dedicata al progetto opensource Kubernetes ed alla tecnologia dei container più in generale tra obiettivi da raggiungere e prospettive future


What would you say to someone who is familiar with Docker and wants to learn more about Kubernetes?

It’s a way to expand your use of Docker across multiple Docker hosts. It’s a relatively new project (as is everything in the Docker world), but it’s based on Google’s long-time experience with running things in containers. Despite those origins, it’s produced by a pretty diverse set of contributors, representing a variety of use cases and perspectives. For getting started with just a regular Docker host, check out this easy how-to.

Are hobbyists developing Kubernetes and related technologies, or does it tend to be larger organizations with itches to scratch?

I haven’t looked at this definitively. I know that the use case is applicable for hobbyist-sized needs—applications of many shapes and sizes benefit from running across multiple machines. I work for a large company, but the use case I’m talking about at SCaLE, the running of a test lab, isn’t limited to a large organization.

Kubernetes is designed to assist in the deployment and scaling of containerized applications across a cluster of environments. How has this helped the proliferation of the container movement?

Docker’s imagery of an application in a shipping container is powerful and attractive, but a container isn’t going to make it across the ocean on its own. If the container movement is not only to blossom but to take root, we need to offload these apps from our developer laptops into production systems. That is what Kubernetes is helping to enable.

What workloads do you see being moved to architectures like Kubernetes in the future?

I think the sky’s the limit. That’s part of what I’m talking about at SCaLE—things like distributed storage and virtualization that you might not expect to run in containers, but that you do want to run reliably and flexibly across a set of machines.

Kubernetes can run on such a large variety of platforms. How much of a challenge is it to maintain that?

This is where a diverse set of contributors comes in handy. Different contributors care about different platforms and use cases, and these separate efforts combine to serve a variety of platforms.

Do you have an automated setup for testing in all these different environments?

I don’t test all the different environments—I typically use an Atomic host fromFedora or CentOS, and use Ansible and Vagrant to automate my cluster creation (learn more here).

How do you see the development of Kubernetes continuing over the next few years? Any major things to support or develop?

In the short term, there are some issues around SELinux integration and running Kubernetes components themselves in containers that I’m keen to see ironed out. I’m also interested in seeing work continue around supporting multiple containerization engines and working alongside other resource management projects (like Mesos) where there’s a good fit.


Link diretto alla fonte