Extending the deployment

Once different services that construct the control plane have been identified, it is a good time to update the initial design draft. An ideal way to start deployment with the upcoming production stage in mind is to assign each physical node a set of roles to run different OpenStack services. That brings us to the following configuration:

• Cloud controller node: This runs the control plane services including OpenStack APIs, identity, imaging, file sharing, telemetry, and dashboard services. Shared services, including a database with a MySQL Galera cluster, the RabbitMQ messaging service, and Memcached, will be part of the assigned cloud controller node role.
• Compute node: This runs KVM as the main hypervisor with Nova compute and networking agent services.
• Network node: This runs Neutron agent services, including layer3, lbaas, dhcp, and plugin agents.
• Storage node: This runs Cinder volume services interfacing with the Cinder scheduler to...