Replicated Embedded Kubernetes
Understanding the Replicated Embedded Kubernetes Cluster
What is Embedded Kubernetes?
Some applications require Kubernetes - either because of architecture decisions or for access to Kubernetes services like CoreDNS or RBAC. Other applications work best on Kubernetes because the hosted, multi-tenant version of the application is running on Kubernetes, and therefore it's the tested, validated and supportable platform.
Replicated's Embedded Kubernetes platform is a portable, predictable and supportable installation of Kubernetes that is crafted from commonly used upstream components. It's designed to be embedded right into an application distribution. An entire Embedded Kubernetes installation can be included in a Replicated airgap package to be run offline, and includes the same types of services expected when launching a managed Kubernetes cluster on a cloud provider.
Kubernetes from Kubeadm
To start, the Embedded Kubernetes cluster is bootstrapped and managed using Kubeadm.
-
One-line Installation
No knowledge of how to bootstrap and provision Kubernetes is required. Kubeadm manages setting up the master and all nodes, incorporating best practices into the setup.
-
Preflight checks enabled
An extensive list of built-in preflight checks will be run to ensure the target system will successfully be able to run Kubernetes.
-
Cloud Agnostic (and even bare metal)
Kubeadm can provision a cluster on any cloud provider, virtualized infrastructure, or even bare metal servers.
Rook and Ceph Storage
The Replicated Embedded Kubernetes cluster is installed with Rook and Ceph, and the installer manages the state of the system.
-
Highly available storage
Rook provisions and operates Ceph, a battle tested, mature system for distributed fault-tolerant storage.
-
Relocatable Volumes
Automatic support for persistent volumes, which enables Kubernetes to reschedule Pods anywhere in the cluster while ensuring data moves with your workloads.
-
Block and Blob Store
Rook provides both block store as a drop-in replacement for standard HostPath volumes, and blob store to implement the S3 protocol on any cluster.
-
CoreDNS Support
By default, Kubernetes will register service and pods with CoreDNS, allowing for container to container communication on the overlay network.
-
Easier On Firewalls
Internal communication can stay inside the cluster, no need to document and require cluster operators to open firewall ports.
Weave Networking
Every Replicated Embedded Kubernetes cluster comes with secure-by-default Weave overlay networking for inter-container communication
Contour Ingress
Contour is a portable ingress controller from Heptio.
-
Built On Envoy
Built on the mature, battle-tested Envoy proxy that powers dynamic infrastructure at massive scale.
-
Cloud native, on prem
Instant reconfiguration, dynamic ingress routes, and native WebSocket support
-
Next Generation Ingress
Leverage next-generation Ingress features like route inheritance and upstream health checking.