Openstack and IDR Ansible Playbooks

The main production IDR consists of three servers:

• Database: A dedicated PostgreSQL server.
• OMERO: OMERO.server and OMERO.web with a highly customised configuration optimised for the data access patterns of the IDR.
• Web proxy: Front-end Nginx proxy that mediates all public access to the IDR. It incorporates an aggressive caching configuration to reduce the load on OMERO.

In addition to the production IDR we also have a virtual analysis platform based around JupyterHub and Docker Swarm. This uses a separate copy of the IDR to ensure that heavy access loads resulting from analysis workflows do not have a detrimental impact on the production server, and requires three or more servers:

• Database
• OMERO: API access only, OMERO.web is currently installed by default but is not required
• Docker manager: The central controlling node for a Docker Swarm running JupyterHub.
• Docker workers: In addition to the Docker manager zero or more Docker workers can be included in the analysis platform. If multiple users are logged on to JupyterHub they should be automatically spread amongst all Docker servers.

The IDR deployment GitHub repository contains Ansible playbooks and full instructions for provisioning resources and deploying the IDR, as well as our internal operating procedures.