The latest version of Kubernetes announced today at the KubeCon + CloudNativeCon 2017 conference makes generally available some key functionalities that promise to make Kubernetes easier to manage, in addition to making the container orchestration engine now widely available on Windows.
Kubernetes 1.9 includes an Apps/V1 Workloads API, which pulls together DaemonSet, Deployment, ReplicaSet and StatefulSet functions for both long-running stateless and stateful workloads. The Cloud Native Computing Foundation (CNCF), which oversees Kubernetes, is also working on a Batch Workloads API that is in beta.
Aparna Sinha, group product manager for Kubernetes at Google, says these APIs collectively mean that IT administrators can now create, for example, a YAML file that they know will work across future releases of Kubernetes.
In all, there are 40 new features being adding to Kubernetes 1.9, including an alpha release of a Container Storage Interface (CSI) that promises to make it easier to add persistent forms of storage to a Kubernetes cluster.
Even though Kubernetes remains a rapid work in progress, use of Kubernetes is rising. A survey of 550 developers and IT professionals published this week by the CNCF finds that 75 percent of the respondents are now using containers in a production environment. Furthermore, 49 percent report they have over 250 containers in use. A full 77 percent of the respondents says they are using Kubernetes.
Much of that progress can be attributed to the fact that in the last three to six months Kubernetes clusters have become easier to deploy. In fact, the next round of development focus for the CNCF is not so much on the core Kubernetes cluster as it is add-on modules of software. For example, the CNCF and Google are working on controllers that will make it easier to spin up Kubernetes clusters that are packaged extensions as well as support for a security and identity API. Sinha says the CNCF is also focusing on tighter integration with Istio, a service mesh announced earlier this year by Google, IBM and Lyft, and the Open Service Broker API developed by the Cloud Foundry Foundation (CFF). The CNCF is also evaluating several serverless computing frameworks as add-on modules to Kubernetes. However, the CNCF does intend to support multiple service mesh technologies on top of Kubernetes.
Sinha says as far as core Kubernetes is concerned, the next two major areas of focus will be adding support for multitenancy and making the boundaries stronger between containers running on Kubernetes. Despite a cadence of upgrades that come every quarter, the CNCF is not at this time considering putting in place a long-term release edition of the cluster platform because much of the upgrade process has been automated.
In general, the sheer weight of the number of vendors contributing to Kubernetes is starting to tell. Usage of Kubernetes already dwarfs rival container orchestration engines. That doesn’t mean rivals will disappear. But the backers of those engines have signaled they will support Kubernetes within their platforms as well. The primary issue now will be finding ways to integrate platforms built on Kubernetes with legacy systems that in many cases are decades old.