spectral cloudthe Bay Area-based modern infrastructure startup, announcement the availability of Palette Edge – a comprehensive platform for deploying and managing Kubernetes at scale across multiple edge computing environments.
Container ship
Pixabay
The rise of containers has prompted cloud providers to offer managed Kubernetes as a Service (KaaS) offerings. Services like Google Kubernetes Engine, Amazon Elastic Kubernetes Service, Azure Kubernetes Service handle most of the heavy lifting related to cluster provisioning, configuration and management. Customers using KaaS offerings focus on the workload while letting the cloud provider manage the infrastructure.
Move to the periphery and customers are alone. They must choose everything between the operating system, Kubernetes distribution, storage, network, service mesh and observability components. Assembling the stack by selecting the right component from dozens of open source and commercial software choices becomes a daunting task. Customers quickly realize that there are too many moving parts in this pile. Selecting one version of an open source component over another can make or break the entire infrastructure stack.
Even after getting the right mix of cloud-native edge stack components, customers often struggle with infrastructure day 2 operations. This includes upgrading and patching the operating system, Kubernetes distribution, and other cluster building blocks. The main challenge here is to perform maintenance at scale, impacting tens of thousands of clusters running in a highly distributed environment.
When running Kubernetes in the cloud or in an enterprise data center, customers delegate the above decisions and operations to the cloud provider or platform provider. When it comes to the edge, there is no managed service or KaaS offering that promises to extend cloud flexibility to the edge. The only exception is signing up for a combination of hardware/software services provided through appliances such as AWS Outposts and Azure Stack HCI. Customers using traditional or custom OEM hardware configuration will need to manage end-to-end management of hardware, infrastructure software, and workloads.
Spectro Cloud’s Palette Edge platform promises to bring the power of managed Kubernetes to the edge. It tackles two important challenges: composing the cluster infrastructure based on the best OSS/commercial components of the native cloud ecosystem and running cluster operations and day-2 workloads at scale. .
The palette has two layers: the centralized management plane and the data plane running at each edge location. The management plane provides the API and user experience to define reusable cluster profiles. The cluster profile is a declarative approach to defining the stack based on the choice of Kubernetes distribution, container runtime, cloud native storage, and network. Additional components such as service mesh, observability and CI/CD software can be easily defined in the cluster profile.
The data plane runs an agent that acts as a proxy for the Palette running at the edge. The agent must be installed on an edge device that establishes the connection between the management plane and the local environment. Clients can pass the cluster definition to the agent, translating it into one or more clusters. The agent is responsible for deploying upgrades and managing patches.
Cluster Profile for Edge
spectral cloud
With Palette’s centralized management plane, customers can take control of tens of thousands of clusters to manage Day 2 operations at scale. The palette effectively becomes the meta control plane leveraging the control plane of each edge cluster.
Palette Edge is built on real metal – a bare metal platform from Spectro Cloud to run cloud-native software. Real Metal is tightly integrated with the popular Metal as a Service (MaaS) from Canonical. Spectro Cloud has extended the open source Kubernetes Cluster API to treat MaaS like any other cloud provider to manage the infrastructure lifecycle.
Palette Edge can run Kubernetes infrastructure in native bare metal mode or in virtualized mode. Natively, it runs the containerized version of the entire stack without the overhead of virtualization. For environments that run a hypervisor, Kubernetes infrastructure can run in a set of virtual machines.
With Palette Edge, Spectro Cloud solves the problem of deploying and managing Kubernetes at scale at the edge. It promises to extend the power and flexibility of the cloud to the edge with a managed Kubernetes as a Service offering.