Start with the state you want
A cluster is a group of machines. You give Kubernetes a description of your application, including its container image and the number of instances you want running. It uses available resources and your placement requirements to decide where new instances can start.
More visitors do not automatically mean more instances. Scaling with demand requires suitable configuration and metrics. Additional machines also need to be provisioned.
Pods, deployments, and services
A pod is the smallest unit Kubernetes schedules. It contains one or more closely related containers. A deployment describes the desired number of pods and how updates should proceed. A service provides a stable address within the cluster for matching pods, even as individual pods are replaced.
Kubernetes can restart crashed containers and replace failed pods. It cannot repair a bug in your application or recover lost data. Storage and backups need their own plan.
Someone still operates the cluster
If you run the cluster yourself, your team maintains the machines, updates, networking, and access permissions. A managed cloud offering takes over some of that work. The division of responsibility depends on the service you choose. Application behaviour, resource requirements, and data storage still need attention.
Using a cluster with MainPath
In MainPath, clusters belong to the organization and projects select one under Hosting. With cloud credentials in place, MainPath can create a managed cluster in your own AWS, Azure, Google Cloud, or Open Telekom Cloud account. You can also import access to an existing cluster through a kubeconfig file.
Imported clusters can host applications, but MainPath cannot use them to provision a managed cloud database. That option requires a cluster MainPath created in a cloud account; classic servers also support the database selection. See Kubernetes clusters and Cloud accounts for the setup steps and requirements.
For the underlying concepts, see the Kubernetes documentation on deployments and recovery after failures.