1. Workloads in GKE

Kubernetes Workload Management Overview

To understand how Kubernetes operates and deploys containerized applications, it’s essential to grasp its core concepts, particularly how Kubernetes expects you to describe workloads and how you can interact with it.

Key Concepts

• Kubernetes Object Model:

  • Kubernetes uses an object model to represent your system’s desired state. Each object is a record of intent, describing the desired state of some component of your system (e.g., applications, jobs, or load balancing).
  • Common Kubernetes objects include:
    • Pods: The smallest deployable units, which represent one or more containers.
    • Deployments: Manage stateless applications and handle scaling, updates, and rollbacks.
    • Services: Expose your applications to other services or external traffic.
    • ConfigMaps and Secrets: Store configuration data and sensitive information.

• Declarative Management:

  • Kubernetes relies on declarative configuration, where you define the desired state of your applications, and Kubernetes works to ensure the actual state matches the desired state.
  • You provide this information through configuration files, typically written in YAML or JSON format.
  • This approach minimizes human errors and reduces the need for manual intervention. Kubernetes automatically ensures that your applications behave as intended, even in the event of failures.

• Kubernetes Components:

  • Control Plane: Manages the Kubernetes cluster, handling the scheduling and orchestration of pods. Components include:
    • kube-apiserver: The API server, which is the entry point for all administrative tasks.
    • etcd: The key-value store that holds cluster configuration data.
    • kube-scheduler: Decides which node should run a newly created pod.
    • kube-controller-manager: Manages control loops that regulate the system’s state.
  • Nodes: Worker machines that run the application workloads (pods) and interact with the control plane.

• Google Kubernetes Engine Concepts:

  • GKE offers two modes:
    • Autopilot Mode: Simplifies cluster management by handling most of the infrastructure setup and configuration. This mode is great for users who want a hands-off experience.
    • Standard Mode: Provides more control over cluster configuration and node management, suitable for users who need to fine-tune their clusters.

• Kubernetes Object Management

  • Kubernetes objects can be managed in several ways:
    • Imperative Commands: These allow you to perform actions immediately via the Kubernetes CLI (e.g., kubectl apply, kubectl delete). This is typically used for quick fixes or testing.
    • Declarative Files: Using YAML/JSON files to describe your workloads and desired states allows for more stable, repeatable, and scalable management of Kubernetes environments.
  • You’ll use tools like kubectl, the Kubernetes CLI, to interact with your clusters and manage workloads.

Google Kubernetes Engine Concepts

• GKE offers two modes:
  • Autopilot Mode: Simplifies cluster management by handling most of the infrastructure setup and configuration. This mode is great for users who want a hands-off experience.
  • Standard Mode: Provides more control over cluster configuration and node management, suitable for users who need to fine-tune their clusters.