Containerization

This page explains the theory behind the containerization.

Container is a method of delivering applications to production.

Container image bundles an application along with its runtimes, libraries, and dependencies together.

Container runtime is the tool that allows to start containers, there are following popular options:

Runtime	Type/Level	Primary Use Case(s)	Key Feature(s)	Notes
Docker Engine	High-Level Platform	Development, end-to-end container workflows	Integrated image building, user-friendly CLI, large ecosystem.	Built on top of containerd and runc.
containerd	Mid-Level Runtime	Container lifecycle management (used by Kubernetes)	Manages image transfer, storage, and container execution.	CNCF graduated project. Used as the default runtime for modern Kubernetes.
CRI-O	High-Level Runtime	Optimized for Kubernetes	Implements the Kubernetes Container Runtime Interface (CRI) cleanly.	Minimalistic, focuses only on being a Kubernetes runtime.
runc	Low-Level Runtime	Spawning and running containers	Implements the OCI Runtime Specification.	The fundamental, process-level component used by almost all other runtimes.
Podman	High-Level Platform	Daemonless container management, Docker replacement	Daemonless architecture, rootless mode, native pod management.	CLI is largely compatible with Docker’s.
Kata Containers	Sandboxed/Virtualized	Enhanced security and isolation	Runs containers within lightweight Virtual Machines (VMs).	Provides stronger separation from the host kernel.
gVisor (runsc)	Sandboxed Runtime	Enhanced security	Intercepts system calls using an application kernel (runsc).	A secure layer between the container and the host kernel.

Container orchestrator is a tool that groups containers together and meets the following requirements:

• Fault-tolerance.

• On-demand scalability.

• Optimal resource allocation.

• Autodiscovery: the ability to identify one another.

• Accessibility from the outside world.

• Seamless update rollbacks without the downtime.

There are popular orchestration tools:

• Amazon Elastic Container Service.

• Azure Container Instances.

• Kubernetes.

• Marathon.

• Nomad.

Most of them are able to:

• Group hosts together while creating a cluster, in order to leverage the benefits of distributed systems.

• Schedule containers to run on hosts in the cluster based on resources awailability.

• Enable containers to communicate with each other, regardless of the host on which they are deployed.

• Bind the containers and storage resources.

• Group sets of containers and bind the to load-balancing construct.

• Manage and optimize the resources usage.

• Allow for implementation of policies to secure access to applications running inside containers.

Docker

Is the most popular tool for defining images and running containers. Check out the corresponding page.

Kubernetes

Kubernetes (kuber, k8s) is a container orchestration tool devoloped by Google based on their internal tool Borg.

Features supported by k8s are:

• Automatic bin packing: Kubernetes automatically schedules containers based on resource needs and constraints to maxmize utilization.

• Extensibility: Kubernetes can be extended with custom features without modyfying the source code.

• Self-healing: Kubernetes automatically replaces and reschedules containers from failed nodes.

• Horizontal scaling: Kubernetes scales applications manually or automatically based on CPU or custom metrics.

• Service discovery and load balancing: Containers receive IP address from Kubernetes, while it assigns a DNS to a set of caontainers.

• Automated rollouts and rollbacks: kubernetes seamlessly rechecks the services and updates them it if needed preventing downtime.

• Secret and configuration management.

• Storage orchestration: k8s automatically ads more storage to the containers from the defined sources.

• Batch execution: k8s supports batch execution, long-running jobs, and replaces failed containers.

Check more in the Kubernetes page.