Hosting Kubernetes Applications
Utilize a single platform for the deployment and management of various workloads ranging from batch jobs and continuous integration/continuous delivery (CI/CD) environments to microservices applications. Replace failing containers as they occur, dynamically scale CPU usage up/down as necessary, all without changing upstream source code!
Kubernetes clusters feature a master server that gives work instructions to servers called nodes, each running its own container runtime such as Docker or RKT to run assigned workloads and communicate with master components for more work assignments. Nodes also create or destroy container instances, adjust networking rules as necessary and monitor performance to make decisions regarding their workloads.
A pod is an application container grouping with similar resource requirements that is deployed onto one node within its cluster and uses a replication controller that defines and manages identical replicas to facilitate container deployment and scaling for developers.
Each container in a pod can create shared storage volumes that are used by all of its containers to share data among applications, providing an effective means for sharing between applications. Furthermore, pods can create network proxies to route requests to services within their cluster.
Kubernetes is cloud-agnostic and can be deployed across AWS, GCP or Azure environments for greater workload portability and reduced data center costs. Third-party tools like kubectl, kind, minikube and kubeadm are available for managing Kubernetes environments.