When running a highly available EC infrastructure, it is necessary to have a way of managing deployed services.
Charm uses multiple Kubernetes clusters managed through a Rancher management console for our development, production and management environments.
Staff Summaries
Member A
Docker allows an application to be placed in a container and run, without worrying about what the host OS is.
Kubernates expands on Docker by managing docker images. Kubernates can monitor the execution of multiple containers and can re-start automatically. If an image is updated, Kubernetes can restart all containers with the new image. (Kubenetes can run images different to Docker).
Kubernetes is clustered. If your service (say a web server) is under a heavy load, add another node to the cluster and Kubernates can seamessly integrate a new container and handle the load bearing. Therefore the whole service can be scaled. Similarly containers can be spooled down automatically under light loads to save on CPU usage.
Ultimately, if a system consists of machines, storage systems and networking, Kubenetes abstracts all of these away from the physical implementation.
Being clustered, Kubermetes relies on 2 (3?) or more nodes to run. One node is designated the Master, accessible only by operators. The others are slaves accessible by the users.
Master node contains system configuration information, a scheduler, and a controller. The controller constantly scans the rest of the cluster and strives to keep it consistent with the configuration.
The scheduler monitors resource use on each node and decides which node can host a new resource if requested by the controller.
Slave nodes are where the containers are deployed, either on a virtual machine or a physical machine.
Member B
- System to manage applications/containers
- When an application is executed, it gives us to be able to scale, deploy and so on
- It can work on various environment(OS, storage)
- Service {Pods {Containers}}
Member C
Kubernetes is a platform for automating container management. You can mount storage, rollout / rollback, and repair failed containers automatically. It consists of clusters, pods, and node services that can be managed vertically and across.