Kubernetes from the User’s Perspective
A central focus of this year’s KubeCon+CloudNativeCon has been on sharing the experience of the end users of Kubernetes. The Cloud Native Computing Foundation has a renewed focus on the enterprises and web-scale companies that have found much value in scalable cloud native technologies. Such users can provide valuable insights to their peers and even contribute to the upstream codebases. So, naturally, many of the most fascinating talks at this year’s event were from engineers and developers who are using Kubernetes and associated cloud native tools in full production settings.
Perhaps most surprising, for us, was a keynote talk given by Apple Software Engineer Alena Prokharchyk. We had heard of Apple using the open source orchestration engine, but had no idea to the extent of the planned usage by the consumer hardware giant. In short, Apple will run the majority of its workloads on Kubernetes clusters.
The company’s developers, she said, “want to adopt cloud native tools for better debugging, logging, monitoring and tracing of their apps. Our responsibility as platform developers is to provide a scalable orchestration layer with secure resource isolation and reliable scheduling.”
Other CNCF technologies are also being used in highly-scalable environments. Another revelation from the show this week is that Slack is now using CNCF’s Vitess to manage over 99% of the messages and other associated data rolling through its system. Vitess provides a way to spread a MySQL database system across multiple servers, in order to vastly improve the scalability and responsiveness of the database as a whole. In this case, Vitess broke Slack’s gigantic MySQL database into 1,000 shards, each responding to user requests with no downtime.
Making Kubernetes easier to digest by developers is a major focus across these organizations. In another talk, Adobe’s Colin Murphy described how this company used the Kubernetes API to build out developer workflows. This approach allows devs to deploy apps to multiple cloud providers or data centers on Kubernetes at scale. Adobe even developed an open source client — written in Go — to help app devs deploy directly to production without learning Kubernetes at all.
Although KubeCon wraps on Friday, The New Stack will be crawling over the wealth of information we picked up during our time there for weeks to come. So check back often at TheNewStack.io to read more great stories about the growing adoption of Kubernetes and other cloud native tools.