As the adoption of OpenStack increases, the industry seeks to learn about OpenStack’s reliability and scalability for production environments. Common questions that typically arise are “How much hardware do I need?”, “Is my hardware working as intended?”, “What are the best practices for scaling out?” among others. The purpose of this reference architecture is to answer these questions by utilizing common benchmark workloads to simulate load upon a scaling OpenStack environment to determine bottlenecks that may arise and how to overcome them. Specifically, this reference architecture shares sample results from Rally used to create boot-storms and validate max-guest per host of the compute servers, tuning around issues found during boot-storm testing, tuning networking MTU sizes, using the CBTOOL to synchronize benchmark results and performance differences when moving from ephemeral storage to cinder utilizing external Ceph servers. It is best suited for system, storage, and OpenStack administrators deploying Red Hat OpenStack Platform 6 with the intent of scaling their private cloud environment.