A 'Pod' in a Kubernetes cluster is the smallest and simplest deployable object. A pod contains one or more containers (e.g., Docker containers). Containers within a pod share the same storage and network resources, such as IP addresses and network ports. A pod also includes a specification of how to run the containers.
A pod is like a self-contained and isolated logical host that includes all the necessary components to run a specific application.
Pods are intended to run single instances of an application, so there are no multiple instances of your application within a single pod. Instead, multiple identical pods (also called 'replicas') are created to run your application multiple times. This is also how you scale up your setup in terms of capacity.
A set of replicated pods are created and managed by a controller, such as a 'Deployment'. Controllers manage the lifecycle of pods and can adjust horizontal scalability. In other words, the controller can adjust the number of pods as needed and does so periodically. Pods are not meant to persist, and they are regularly replaced, for example during an update. If a pod fails for any reason, it will not come back, but will be removed and replaced by a new one.
Pods are executed on nodes within your cluster. Once created, a pod remains on a node until it is removed for any reason, such as when a node no longer has resources (e.g., RAM) available or when the node crashes.