ReplicaSet

A higher-level Kubernetes controller that manages ReplicaSets to provide declarative updates for Pods. A Deployment describes a desired state, and the Deployment controller changes the actual state to the desired state at a controlled rate.

• Key Function: Manages the rollout and rollback of application versions by creating and scaling underlying ReplicaSets.
• Primary Use Case: The standard controller for deploying stateless applications, enabling features like rolling updates and rollbacks that a bare ReplicaSet does not provide.
• Relationship to ReplicaSet: When you update a Deployment's Pod template, it creates a new ReplicaSet and scales it up while scaling the old one down.

A Kubernetes workload controller used for managing stateful applications that require stable, unique network identifiers, persistent storage, and ordered, graceful deployment and scaling.

• Key Difference from ReplicaSet: Provides stable, persistent identity for each Pod (e.g., web-0, web-1). Pods are not treated as interchangeable commodities.
• Primary Use Case: Databases (e.g., MySQL, Cassandra), distributed data stores, and any application where Pods have a unique role or persistent data that must survive rescheduling.
• Scaling Behavior: Scaling down occurs in reverse order, and each Pod's persistent storage is not deleted unless the StatefulSet is explicitly destroyed.

A Kubernetes controller that ensures a copy of a Pod runs on all (or a subset of) Nodes in the cluster. It is used for cluster-level services that must run on every node.

• Key Difference from ReplicaSet: Pod placement is node-centric, not count-centric. It targets nodes based on their labels, not a desired replica count.
• Primary Use Cases: Running cluster storage daemons (e.g., glusterd, ceph), log collectors (e.g., fluentd, logstash), and node monitoring agents (e.g., Prometheus Node Exporter).
• Automatic Scaling: Automatically adds a Pod to new nodes that join the cluster and removes Pods from nodes that are removed.

A Kubernetes controller that automatically scales the number of Pods in a workload resource (like a Deployment or ReplicaSet) based on observed CPU utilization, memory consumption, or custom metrics.

• Key Function: Dynamically adjusts the spec.replicas field of a target resource to match observed demand.
• Primary Use Case: Automating scalability for applications with variable load, ensuring performance during peaks and cost-efficiency during troughs.
• Interaction with ReplicaSet: The HPA continuously monitors metrics and updates the replica count on the targeted ReplicaSet or Deployment controller, which then takes action to match the new desired state.

A Kubernetes policy that limits the number of concurrent voluntary disruptions to Pods belonging to an application, ensuring high availability during cluster maintenance operations.

• Key Function: Defines minimum availability (minAvailable) or maximum unavailability (maxUnavailable) for Pods managed by a ReplicaSet, Deployment, or StatefulSet.
• Primary Use Case: Protecting application availability during voluntary disruptions like node drains, cluster upgrades, or manual Pod evictions.
• How it Works: The Kubernetes cluster components (e.g., kubectl drain) respect the PDB, ensuring they do not evict more Pods than the budget allows, thus guaranteeing a certain number of replicas remain running.

Kubernetes controllers for managing finite, run-to-completion tasks, as opposed to long-running services managed by ReplicaSets.

• Job: Creates one or more Pods and ensures that a specified number of them terminate successfully. Used for batch processing, data analysis, or one-off tasks.
• CronJob: A higher-level controller that manages Jobs on a time-based schedule, similar to a Unix cron job.
• Key Difference from ReplicaSet: Pods are expected to terminate after completing their work. The controller's goal is successful completion, not maintaining a continuous set of identical replicas.
• Use Case: Running database migrations, sending nightly report emails, or performing scheduled backups.