Managing Kubernetes Clusters for Platform Engineers
Managing Kubernetes cluster gets tricky for platform engineers, but reusability and scalability have become more feasible.
Spectro Cloud has the notion of a cluster profile in a UI that embodies repeatability so a user can apply a configuration in any cloud, anywhere, but still keep the same profile, whether for the cluster, the application, or a virtual cluster.
Spectro Cloud simplifies the process through its Palette service, which allows platform engineers to decouple the cluster, Nic Vermandé, Senior DevRel Manager at Spectro Cloud said. Palette provides a platform for managing Kubernetes environments in the cloud or a data center.
Virtual clusters distinguish the Palette service. In the demo, Vermandé explained how to provision the Kubernetes cluster from the existing cluster in the Palette environment. In the workflow, platform engineers create a new cluster with Palette in the cloud or on-premise. Out of that, developers receive virtual clusters to deploy the application in the cloud or on-premise.
The actual cluster in Palette uses Loft Vcluster, which guarantees isolation. Vcluster is an open source project that allows engineers to spin up lightweight, virtual Kubernetes clusters that run inside the namespaces of an underlying Kubernetes cluster. In doing so, Vcluster provides a more resource-efficient manner of spinning up multiple clusters and provides a form of soft multitenancy without handing out administrator credentials to more users than needed, writes Mike Melanson in The New Stack.
By using virtual clusters, Palette offers a harder form of isolation than softer forms such as Kubernetes namespaces, Vermandé said.
The demo shows how Palette allows developers to deploy applications to a sandbox environment. Based on the same concept as Palette provides, developers get an application profile for defining a Kubernetes cluster, but this time in terms of defining the app declaratively and then deploying it to their sandbox.
The demo shows what fundamentally defines how Spectro Cloud serves the platform engineer by providing an environment to decouple the cluster targets from its configuration.
For example, users may set up a profile for a cloud service such as Google Cloud, Vermandé said. A user may add layers to the cluster and then assign the particulars, such as the region for deployment, with features for day-to-day management.
A user may see all the various workloads and deployment pods. They may plan to scan the cluster. They can include pen testing and different other capabilities.
The concept behind Palette is to manage day-to-day operations, from deploying the cluster to managing the cluster to creating isolation and deploying applications. It uses the same methodology, the same language, and open source-based software that already exists, Vermandé said.
Palette reflects how cloud native technology vendors are leaping in the experience they offer enterprise users. They are managed in a way that serves the customer. For instance, Spectro Cloud engineers contribute to upstream open source projects.
“I would say the key thing here is are two things: reusability, and then scalability,” Vermandé said.