Some of the questions that most intrigue humanity are undoubtedly related to the universe, how it works and what it is made of. And that is precisely what CERN, the European Organization for Nuclear Research, tries to uncover by providing a unique range of particle accelerator facilities to researchers around the world.

Unsurprisingly, this work results in an enormous amount of data — more than half an exabyte — and many interactions with diverse academics who are pushing the frontiers of science and technology. To deal with this, CERN’s technology team has increasingly embraced containerization and cloud native practices, choosing Kubernetes for orchestration.

“CERN is known for having a lot of data and requiring a lot of computing capacity to analyze all this data, but actually we also have a very large community and we have a lot of users and people interested in the stuff we do,” said Ricardo Rocha (pictured), computing engineer at CERN. “We’ve been migrating all our group infrastructure into Kubernetes and, in this case, actually OpenShift, and the challenge there is to run a very large amount of websites on Kubernetes.”

Rocha spoke with John Furrier, host of theCUBE, SiliconANGLE Media’s livestreaming studio, during KubeCon + CloudNativeCon. They discussed CERN’s experience with the large-scale deployment of Kubernetes, as well as how the organization has used machine learning and automation in its processes. (* Disclosure below.)

Machine learning is another focus

The use of Kubernetes has allowed CERN, one of the largest and most respected scientific research centers in the world, to run and manage the lifecycle of more than 1,000 sites, including updates and new deployments. In addition, an operator was developed for this purpose, according to Rocha.

“Compared to the infrastructure we had before, [the key innovation] is this notion that we can develop an operator that will manage a resource, in this case a website,” he said. “And this is something that is not always obvious when people start with Kubernetes. It’s not just an orchestrator; it’s really the API and its capability of managing a huge amount of resources, including custom resources.”

Machine learning has also been a big topic for CERN. A lot of its workloads are migrating to accelerators and can benefit from ML capabilities to improve analytics.

“What we are trying to do is just offer a service to all our users, where we help them with the infrastructure so they don’t have to focus on that [and] they could focus just on their workloads,” Rocha explained.

That way, CERN does everything, from exposing the data systems it has in-house so that everyone can access them to preparing the data and then doing some iteration using netbooks. This includes disability training with a potentially large amount of GPUs and then storing and serving up the models.

“And all of this is managed with a coordinated cluster underneath,” Rocha stated. “We had a lot of knowledge of how to handle Kubernetes and all the features that everyone likes — scalability, the reliability, auto-scaling, [which] is very important for this type of workload. This is key.”

The deployment of Kubernetes and the use of machine learning are the topics that Rocha will explore in his two KubeCon speaking sessions during this week’s event.

Watch the complete video interview below, and be sure to check out more of SiliconANGLE’s and theCUBE’s coverage of KubeCon + CloudNativeCon. (* Disclosure: Cloud Native Computing Foundation sponsored this segment of theCUBE. Neither CNCF nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE