RPCR Architecture

While based on Red Hat Enterprise Linux OpenStack Platform, the architecture of the Rackspace Private Cloud Powered By Red Hat (RPCR) solution is tailored to meet specific customer requirements along with our Fanatical Support. RPCR requires shared storage that uses the Converged Storage service (ceph), and can optionally use a NetApp device. For detailed guidance about architecting Red Hat Enterprise Linux OpenStack Platform implementations, refer to the Architecture Guide at: https://access.redhat.com/documentation/en/red-hat-enterprise-linux-openstack-platform/.

Note that the reference implementation described in this link is not identical to the RPCR implementation due to specific Rackspace customizations and optimizations. Refer to the RPCR detailed feature list in this document for a list of differences, or contact your Rackspace representative.

We can deploy RPCR in your data center or ours. For solutions deployed in a Rackspace data center, the following table describes a minimum proof of concept implementation.

Hardware load balancer2
Firewall appliance2
Converged Storage service (ceph) node5
NetApp appliance (optional)As needed based on storage requirements
Controller node3
Installation node (Can additionally function as a Controller node1
Compute node5
Network node (Nexus switches)2

The architecture of RPCR provides for the following requirements:

Clustered databaseThe database that backs the OpenStack services is clustered and allows for horizontal scaling. Pacemaker provides failure recovery for the database cluster.
Clustered queueThe queuing service is clustered and allows forhorizontal scaling. Pacemaker provides failure recovery for the queue.
Fail-overServices are restarted with minimal interruption.
Horizontal scaling of servicesOpenStack services are scalable up or down across multiple hardware nodes as computing needs change.
Host operating system installationThe host operating system is installed by Red Hat Director. The installation also manages patches, upgrades, and configuration change implementation
Hardware load balancingLoad balancing is provided through hardware appliances. An exception to this is a stand-alone HAProxy device which is treated as a physical load balancer to the cloud.
MonitoringMonitoring in Rackspace data centers is provided using MaaS and is deployed automatically.
Single installation nodeA single hardware node is used to deploy all OpenStack services and to maintain and monitor nodes.

RPCR reference architecture

The following figure illustrates the RPCR general reference architecture:

The following figure illustrates the RPCR network architecture:

The following figure illustrates the general ceph architecture for RPCR: