After adopting the microservice architecture, with the development of the system, there will be more and more microservice applications to be managed. At this time, a mechanism is needed to manage the life cycle of each service (deployment, application, update, recycling, etc.) and many Service orchestration and planning (on which machine to deploy, how much resources it takes, etc.).
Kubernetes integrates the physical machines that can be used to deploy services and abstracts it into a deployment platform, so that no matter how many machines are supported at the bottom, it seems to developers to deploy on one machine.
The hardware of kubernetes is divided into management nodes and working nodes. The management node is used to control and manage the entire system, and the working node is used to run the actual deployed application.
The management node mainly includes four parts:
- Kubernetes API: Users and other components in the system communicate through API
- Scheduler: Decide on which worker node the application is deployed
- Controller Manager: Process system-level events, such as copying an application, recycling failed applications, etc.
- etcd: Distributed data storage, used to record system configuration and other information The
working node contains the following components: - container
- Kuberlet: communicate with the Kubernetes API of the management node and manage the containers in the working node
- Kubernetes Service Proxy / Kube-proxy: communication between proxy components