What it can it do, and how to do it:

Cube in a Box

What it can do
 
The ODC Reference install - Cube in a Box

A distributable, ready to run reference install is available as the “ODC Reference Install”, or Cube in a Box (CIAB). Where the Sandbox install provides an accessible, externally managed platform to trial the features of the Open Data Cube, the Reference Install is designed to provide a ready to run installation of an independent Open Data Cube, on an organization's own resources. This can be locally hosted or installed on the Cloud, but is specifically maintained and customised by the user. The same DCAL applications are available as in the Sandbox install, and the same data is indexed, however the reference install allows the addition of any other data available to the user, be it commercial imagery, derived products, or in-situ data
The Reference install is made available as a Docker Image, which is installed to either a local machine or to a cloud service, such as AWS. A Docker image provides a distributable, easily maintained reference installation, which is common to both the local install and the cloud install. The Reference install is made available as a Docker Image, which is installed to either a local machine or to a cloud service, such as AWS.
Detailed instructions for either implementation are available on the CRC-SI CIAB GitHub, and are included below.

How to do it
 
The Reference Install: Cube in a Box

As described above, the Reference Install is a docker image, maintained on the Open Data Cube GitHub. Installing ODC to a Docker container makes it trivial to install, maintain and redistribute. It also provides flexibility in migrating to a cloud platform or local machine if desired. When installing, it is automatically updated to the most recent version available and will download the newest version of the DCAL.
There are two options to getting started with the Open Data Cube Reference Install "Cube in a Box". The first is using local hardware, such as a desktop pc, laptop or server. The second is installing to a cloud service, such as Amazon Web Services. Both make use of Docker, a freely available platform that simplifies the distribution of ready to run applications.
As a very brief overview of the components to Docker, a Docker image is a read-only template with instructions for creating a Docker container. A Docker container is a runnable instance of an image, which shares resources with its host operating system, as opposed to a Virtual Machine which must run its own OS. More information is available on Docker's extensive documentation.
 

Local Install

If you are unfamiliar with Docker or Jupyter, this guide will take you through downloading Docker to setting up a Cube in a Box Jupyter Server running the Open Data Cube. While these instructions suit both OSX and Windows, the steps can easily be adapted for other a different OS, using the appropriate Docker version. Please note the ODC image, plus additional dependencies installed to the container will come to several hundred Megabytes. However, as satellite imagery are stored on the cloud and only loaded when called, disk space requirements are minimal.

To access the Landsat 8 PDS hosted on AWS, you will need an Amazon Web Services account, and associated AWS access keys. Without them, you will need to provide your own data and index it yourself. USGS ARD is available from this resource page. You can also use your AWS account to launch a cloud based Data Cube on AWS, as detailed below. To get started with AWS, you will need to sign up for an account, then access or generate your secret keys from the Security Credentials management.

The video below provides a walk through of downloading and installing the Cube in a Box. Note that you will first need your AWS keys, and to have installed Docker. The detailed instructions are also available on the Cube in a Box GitHub repository.

Cloud Install

If you are unfamiliar with AWS, this detailed guide can help you set up an AWS account, then create the necessary AWS components to run ODC on the cloud, using the provided template from the Magic Link on the CIAB GitHub.

The cloud install will cost around $10 USD per month for the lowest power server, but provides easy, remote access to users. If it is installed to US-West 2 (Oregon) it will sit next to the Landsat-8 PDS and tile downloads will be very rapid.