Running AEM in Docker

Thinking to run AEM in container ?

The basic questions that strikes the mind

• Docker is designed for microservices
• AEM is a monolith
• Docker runs the best with stateless containers
• AEM has a big filesystem snapshot repository can be of the size of GB's

I started with creating a basic image which contains every thing within the container , one complete image which contains the image, bundles and repository.

Basically we have 2 use cases

1) For lower environments we can have every thing inside one container - Binaries and Content

2) For higher environment (UAT and Production) we need repository persistence

Solving the first use case is easy, use the basic image from the Here

The challenge is to achieve the repository persistence. And we were able to achieve it by using the concept of volumes .

We can use Docker data only containers, in which we create a volume from an image

docker create -v /<path-from-image> --name data-store <<aem_image>> /bin/true

what above command will do, is host the path from the image as a persisting volume as a container, even if we remove the image the volume still remains.

When you start your AEM container just use the command "volumes-from" to mount the repository from the data container

docker run -d --name=aem-author-instance --volumes-from repo-segment-store --volumes-from repo-data-store -p 6602:6602 aem_author_demo

Another way to achieve the repository persistence is using the host volumes

we can mount a directory on the docker host, and when we run the container it picks the path from the mount .

use the command to create the volume

docker create -v /aem/segmentstore name repo-segment-store aem_author /bin/true

and use the command to host the volume when starting the container

docker run -d -P --name aem-author-3 -v </home/ubuntu/datavolume/dockervolume/binarystore/segmentstore>:</adobe/segmentstore>  -p 4502:4502 <test_volume>

So we can achieve both the use case using docker, and the advantages we get are very impressing

Environment setup time for a new person in team or recreating the environment for a urgent defect is just matter or seconds .

In production or UAT , we can spin around a new container on demand .

Taking this concept a little further if we combine this solution with an configuration system like "Ansible" and an orchestrating solutions like "kubernetes" the possibilities are unlimited, we can completely eliminate the issues related to environment and code progression from one environment to other .

In the next article I will post the solution for using AEM with Docker+kubernetes+Ansible , stay tuned .