An application container is a relatively new container type. It is an application, service, or even microservice centric solution that usually runs just a single process inside. As a result, application containers promote the creation of immutable and ephemeral infrastructure. If an application or service needs to be updated, a whole new container is built (with the required adjustments) from the appropriate image. Then, it is provisioned to replace the existing running container instance.
In the early days, such complexity, as well as the lack of proper isolation and incompatibilities between application containers and some technologies (like Java and database runtimes), greatly slowed the adoption of application containers. Developers had to implement significant adjustments to ensure that their technologies and additional functionality could run correctly inside application containers.
Also, as a stateless solution, application containers cannot natively store state information, so such operations are delegated to external persistent storage systems. However, according to the developers of purely stateless containers, bringing state to deployment is an obsolete architectural practice. They assert that the best and cleanest approach for the containers is to perform operations that do not require keeping the state.
Nowadays, many supplementary tools are implemented to make it easier to use services that use application containers. As a result, developers can benefit from the highly specialized units of application containers without the complexity of managing them.
Several application container implementations are actively developed: Docker, CRI-O, containerd, and some others. At the moment, the platform utilizes Docker as the most widely adopted technology for application containers. The platform provides comprehensive deployment automation, fast server provisioning, and stateful data storage and processing tools (sessions, logs, configs, etc.) to make it easy to use application containers.
However, some applications and technologies may still have problems while migrating to application containers (due to lack of proper isolation, statelessness, and single-process requirements). This is especially relevant for the complex application container orchestration tools like Kubernetes. In cases of migration difficulties, system containers can be considered a better-suited alternative for containerization.
