virtual-kubelet/virtual-kubelet
Virtual Kubelet
Virtual Kubelet este o implementare personalizată de kubelet care permite unui control plane Kubernetes să trateze platformele serverless și API-urile de calcul externe ca noduri de cluster standard. Acesta funcționează ca un orchestrator de calcul extern, traducând comenzile standard de orchestrare în apelurile API specifice necesare pentru a rula containere pe infrastructură non-nativă.
Proiectul oferă o interfață pluggable de furnizor de noduri virtuale, permițând dezvoltatorilor să implementeze furnizori de noduri personalizați care mapează evenimentele ciclului de viață al pod-urilor la backend-uri externe specifice. Acest lucru permite emularea comportamentului standard al nodurilor, ceea ce permite controllerului de cluster să programeze workload-uri pe platforme terțe.
Sistemul acoperă orchestrarea calculului extern, programarea resurselor în cloud hibrid și expansiunea capacității clusterului. Include capabilități pentru urmărirea ciclului de viață al pod-urilor, rutarea cererilor administrative pentru log-uri și execuție, precum și colectarea și exportul telemetriei nodurilor și metricilor pod-urilor pentru a susține autoscaling-ul orizontal nativ.
Deployment-ul este suportat prin manageri de pachete folosind bundle-uri de configurare predefinite.
Features
- External Compute Orchestrators - Translates standard Kubernetes orchestration commands into actions on serverless platforms or external compute APIs.
- Virtual Node Providers - Provides a pluggable interface to define how the Kubernetes control plane manages and schedules workloads on external cloud backends.
- Compute Lifecycle Mapping - Translates pod and container lifecycle operations into external API calls to manage compute without virtual machine infrastructure.
- Compute Node Emulation - Simulates standard compute node behavior to enable the cluster controller to schedule workloads on external infrastructure.
- Pod Lifecycle Event Mapping - Allows developers to define the logic required to translate cluster pod events into actions on specific external compute platforms.
- Custom Kubelet Implementations - Provides a custom implementation of the kubelet that allows Kubernetes to treat external services or serverless platforms as cluster nodes.
- External Compute Orchestration - Manages containers on non-native infrastructure by translating Kubernetes pod lifecycles into external API calls.
- External Compute Scheduling - Schedules containers on external APIs or serverless platforms by emulating a standard node to the cluster controller.
- External Node Integration - Connects external services or serverless platforms to a Kubernetes cluster so they function as standard compute nodes.
- External Resource Orchestration - Represents serverless platforms as standard nodes to run and manage containers on non-native infrastructure.
- Kubernetes Node Emulation - Simulates standard node behavior to allow the control plane to manage resources on third party platforms.
- Orchestration API Bridges - Maps Kubernetes container orchestration commands to the proprietary API calls of serverless or external cloud platforms.
- Provider Plugin Architectures - Provides a pluggable interface to decouple Kubernetes pod lifecycle events from service-specific external compute APIs.
- Serverless Kubernetes Integrations - Maps Kubernetes pod lifecycles to external API calls for running containers on non-native compute infrastructure.
- Infrastructure Provider Interfaces - Provides a standardized interface for abstracting and interacting with external cloud and serverless compute APIs.
- Administrative Request Proxies - Proxies administrative requests for logs and execution from the Kubernetes control plane directly to external service providers.
- Administrative Request Routing - Forwards API server requests for logs, interactive execution, and metrics directly from an external compute provider.
- Cluster Expansion - Adds virtual nodes to a cluster to handle burst traffic using external compute without managing VM infrastructure.
- Pod Lifecycle Management - Manages the creation, updating, and deletion of pods while monitoring their status on an external provider.
- Hybrid Cloud Kubernetes Management - Schedules workloads across a mix of traditional virtual machines and external cloud provider services.
- Node Capacity Reporting - Sends node capacity, addresses, and operating system details to the controller to enable accurate scheduling.
- Pod Lifecycle State Tracking - Tracks the status of pods on external platforms and synchronizes that state back to the cluster control plane.
- Per-Node Pod Metric Collectors - Gathers pod-level statistics from virtual nodes to support horizontal autoscaling.
- Pod Metric Exporters - Sends resource statistics for pods to the control plane through a dedicated metrics interface.
- Virtual Resource Capacity Reporting - Reports virtual resource capacity and pod metrics to the scheduler to enable native horizontal autoscaling and placement.
