The Cloud Native Computing Foundation (CNCF) has made the Kubernetes Universal Declarative Operator (KUDO) management toolkit created by D2iQ as a sandbox level project.
D2iQ CTO Ben Hindman says Kudo traces its lineage back to the open source Mesos project. It was created to automate the management of Mesos clusters when D2iQ was known as Mesosphere. Since then D2iQ has been working on an implementation that runs natively on Kubernetes.
The acceptance of KUDO comes on the heels of a move by the CNCF to also accept an Operator Framework from Red Hat as an incubation-level project. That initiative promises to make it easier to create Operator management tools that unify the management of applications and the Kubernetes clusters they run on.
Hindman says KUDO, in contrast, is a toolkit designed for IT teams that need to manage applications after they are deployed on Kubernetes clusters. Operators are increasingly being used in place of low-level YAML files to automate tasks such as running backups, rebalancing data, scaling or changing configurations. KUDO makes it possible for IT teams to first declaratively create an Operator and then manage the life cycle of those Operators.
IT teams are increasingly interested in creating their own Operators for multiple applications that might span a fleet of Kubernetes clusters versus trying to manage multiple Operators from IT vendors that might be designed to automate tasks for a single application, notes Hindman. Regardless of the use case, IT teams that don’t tend to have a lot of programming expertise need to be able to declaratively create their own Operators and employ them as they best see fit, he says.
In the meantime, many IT teams are trying to navigate when they might rely on a Helm Chart to package an application for deployment on a Kubernetes cluster versus employing a set of Operators to both deploy and manage the entire Kubernetes environment.
As Kubernetes clusters become more widely adopted in enterprise IT environments, management issues are coming to a head. A recent CNCF survey found that 81% of the IT teams that had deployed Kubernetes clusters in a production environment had more than 20 machines in their environments. A lack of readily accessible management tools, however, is clearly holding back wider adoption; not every IT organization that wants to deploy containerized applications has a site reliability engineer (SRE) capable of managing Kubernetes clusters using a command-line interface (CLI).
IT teams should expect to see an explosion of Operators being launched in the months ahead, many of which will be redundant as IT teams and vendors alike create them. That could wind up being too much of a good thing. However, in most cases Operators will prove superior to manually managing massive amounts of YAML files that someone in IT would have to make sure are executed in order simply to stand up an application on a Kubernetes cluster.