Kubernetes: Data Center or Edge?
“The Kubernetes cluster, as we know it, is no longer in the data center,” Keith Basil, Rancher vice president of edge solutions, told The New Stack earlier this week. He was discussing how his company’s stripped-down version of Kubernetes distribution, called K3s, was proving to be quite popular, not just as a handy developer tool, but as a production platform for retail stores, fast-food restaurants and industrial platforms. Rancher has submitted K3s to the Cloud Native Computing Community as an incubation project.
There is a definite energy around K3s and edge computing, the kind that settled on Kubernetes itself in its early years, explained Karthik Gaekwad, Verica’s head of cloud native engineering, at a KubeCon press conference on the topic of edge computing.
It was at last year’s KubeCon where Rancher itself was “overwhelmed with the number of use cases that started pouring in,” said Rancher’s Shannon Williams, at the same presser. “More people were coming to talk to me about edge and K3s then data center and traditional hybrid cloud Kubernetes implementations,” Williams said.
But what does Kubernetes offer for the edge computing/Internet of Things/embedded systems world? It’s not so much the scalability that is so desired in the data centers, but a range of other factors, including security and automation, as its promise to not lock the user into a specific, often narrowly understood, vertical platform.
But perhaps most importantly, K8s offers the promise of application portability, Gaekwad noted.
“Kubernetes treats [all] devices in a similar manner. You can bring whatever devices you want, and potentially run Kubernetes, or K3s, on them,” Gaekwad said. The problem of “can you run this code on this device in an easy manner platform?” — as Gaekwad puts it — is solved in a very “elegant manner.”
As Kubernetes development advances, will this bifurcation of markets lead to a forking of Kubernetes for these radically different set of users? Or, like Linux, will it remain robust enough to accommodate both markets?
“We have to be careful with Kubernetes and how it grows,” admitted Frederick Kautz, head of edge infrastructure for Doc.ai, at the press conference. “We are at a crossroads in telecom and edge. We can stick with enterprise and make use of everything that is coming out of that. Or we can split and head off in another direction, and eventually, we’ll be holding the bag with all the complexity that is there, without the size of the community. The best case is that we find a way to coexist.”