With the recent success of DockerCon and the rapid adoption of containers by enterprise, there’s no doubt that the container revolution is here to stay. Now, Google has stepped into the fray to bring containers to scale with Kubernetes, an open-source solution that abstracts container management, allowing developers to focus on services as a whole, rather than focusing on deploying particular containers on particular machines.

In a CrowdChat on July 23rd, 2015, several members of the Google Cloud team answered questions about the development, implementation and future of Kubernetes. CrowdChat participants included Brendan Burns, lead engineer on Kubernetes; Miles Ward, global head of solutions at Google Cloud Platform; Navneet Joneja, product management, Google Cloud Platform; Kit Merker, product manager at Google; and Craig McLuckie, product manager. The CrowdChat hosts included John Furrier, Jeff Frick and Stu Miniman from theCUBE, SiliconANGLE Media’s production team.

Why did Google open source Kubernetes?

Given all the questions that surround open source — business model, enterprise-readiness and quality control — this question prompted a considerable response during the CrowdChat.

Burns emphasized that “the future of Cloud is hybrid,” which means that applications need to be able to run across multiple Clouds and on-premise databases, even during development. Open source is the best way to take advantage of an “evolving ecosystem” and get contributions from across the tech industry.

For McLuckie, the answer was clear: They’ve “seen first hand the power of the community in building better tech.” Joneja added that this is a way to “accelerate that transition” to Cloud-native apps, which better enables developers to take advantage of the unique capabilities of the Cloud.

This sparked a question by Steve Chambers about what, exactly, those unique capabilities are. “It’s not a binary thing, right?” he asked. “More a spectrum? And some enterprises have good Cloud-like capabilities sans Cloud now.”

Joneja responded by saying that it is “absolutely a spectrum, and when those capabilities exist on owned and operated infrastructure it makes even more sense to build apps in a way that allows them to take advantage of them. That transition is happening slowly; we want to accelerate it.”

Is there a typical use case or customer profile?

McLuckie fielded this question by saying that their focus has been on “stateful and stateless web app hosting,” with many customers using it to test scale-outs and even build their DevOps pipelines. They also plan to use it to run scale-out batch workloads, since the platform will enable them to be run on the same infrastructure that will eventually host the app.

EMC’s Brian Gracely asked whether Kubernetes will expand to focus on stateful architectures. McLuckie responded that this is critical. “Inside Google we run all our stateful services on the same clusters as everything else,” he explained. “Even today we have pretty solid volumes support for local and network attached storage. I expect more intelligence around state to come.”

What scale does a company need before Kubernetes will be necessary?

Given that most companies don’t operate at Google scale, is Kubernetes even necessary for most businesses? Miles Ward responded, saying that the other aspects of Kubernetes make it useful for businesses of all sizes.

“Especially with a few small teams working on testing software, kubernetes can be hugely beneficial on even a small cluster,” he said.

Merker agreed, asking, “Who doesn’t want auto-healing?” Several commenters also pointed out that planning for scale makes sense for any application that may one day need it. If large-scale thinking works on the small scale, why not develop that way in the first place? Burns drove this home, saying, “I think once you have users who expect your service to be up all the time, you’re at that scale. A lot of what we do is make it possible to manage and deploy software reliably at any scale.”

See the full CrowdChat below:

photo credit: sensitive noise / obvious 2 via photopin (license)