Introduction
On-premises OpenShift deployments, especially on platforms like vSphere, often grapple with maintaining cluster quorum in the face of data center failures. To address this, OpenShift 4.17 introduces the 3+2 control plane architecture, a five-node configuration designed to enhance resilience and simplify deployment in multi-failure-domain environments.
Challenges with Traditional Three-Node Control Planes
Deploying a three-node control plane across two data centers can be problematic. If one data center becomes unavailable, the cluster loses quorum, leading to potential downtime. To mitigate this, organizations often deploy the third control plane node in a separate location, which can be logistically complex and costly.
Enhanced Control Plane Configurations in OpenShift 4.17
OpenShift 4.17 addresses these challenges by introducing the option to scale control planes up to five nodes on bare metal platforms. This configuration, known as 3+2, involves deploying three control plane nodes in one failure doamin and two in another. This setup enhances fault tolerance and ensures cluster availability even if an entire data center fails.
According to Red Hat’s announcement, “Organizations often deploy OpenShift Virtualization stretched across two failure domains using a 3+2 control plane, or 5-nodes control plane configuration. The 3+2 configuration strikes a balance between providing high availability and keeping the cluster architecture relatively simple to manage. If there is a domain failure in a 3+2 configuration, cluster stability is preserved. Specifically, if you lose the 2-node failure domain, quorum is retained. If you lose the 3-node failure domain, your cluster is still operational with the remaining 2-node control plane, however the cluster is now in a state with reduced redundancy and cannot tolerate another control plane node failure without losing quorum.”
This new design benefits not only OpenShift Virtualization but also any StatefulSet deployments.
Benefits of the 3+2 Control Plane Configuration
- Improved Fault Tolerance: The cluster can withstand the loss of an entire data center, maintaining operational continuity.
- Simplified Deployment: By eliminating the need for a third, separate location for the control plane node, deployment becomes more straightforward and cost-effective.
- Enhanced Support for Stateful Workloads: This configuration is particularly beneficial for stateful applications, such as those using RabbitMQ, as it provides a more robust and reliable environment.
Implementing the 3+2 Control Plane Configuration
To implement this architecture, ensure that your OpenShift 4.17 deployment is on a bare metal platform, as this configuration is currently supported only in such environments. Carefully plan the distribution of control plane nodes across your data centers to achieve the desired fault tolerance.
Conclusion
The introduction of the 3+2 control plane configuration in OpenShift 4.17 offers a significant advancement for on-premises deployments, particularly those utilizing vSphere. By providing enhanced fault tolerance and simplifying deployment architectures, this feature enables organizations to maintain high availability and resilience in their OpenShift clusters, even in the face of data center failures.
Further Reading: What You Need to Know About Red Hat OpenShift 4.17
https://www.redhat.com/en/blog/what-you-need-to-know-red-hat-openshift-417
Test scenario
In a simple test scenario, only two nodes of the five-node RabbitMQ cluster were operational, yet the RabbitMQ endpoint remained reachable. Messages could still be published and consumed without issues, demonstrating the robustness of the 3+2 configuration for stateful workloads.
