Red Hat and Cloudera publish a 100-VM OpenShift Virtualization blueprint

Red Hat has published a deployment blueprint for running Cloudera Data Platform on OpenShift Virtualization, turning an earlier partner validation into a concrete reference architecture for platform teams. The design was tested with Cloudera Base spread across 100 Red Hat Enterprise Linux virtual machines and a separate bare-metal OpenShift cluster for containerized data services, according to the Red Hat engineering post and Cloudera’s validation summary.

What the teams validated

The larger host cluster used 110 physical servers: three OpenShift control-plane nodes, 24 storage/worker nodes and 83 additional workers. Cloudera Base ran in 100 RHEL virtual machines, including three master nodes, 95 workers, one Cloudera Manager node and one utility node. A second, seven-node bare-metal OpenShift cluster hosted Cloudera Data Services, with three control-plane nodes and four storage/worker nodes.

Cloudera’s published test matrix identifies OpenShift and OpenShift Virtualization 4.17.42, RHEL 9.5, Cloudera Manager 7.13.1 CHF4 or later, Cloudera Base 7.3.1.400 SP2 and Data Services 1.5.5. That version specificity matters: this is evidence for a tested combination, not a blanket compatibility statement. Red Hat directs readers to Cloudera’s support matrix before choosing current product versions.

Functional validation covered Cloudera Data Warehouse queries, Spark jobs in Cloudera Data Engineering and AI Workbench sessions in Cloudera AI. The teams also exercised a bank-branch analytics pipeline from ingestion through processing, query and visualization. Identity and data controls in the test included FreeIPA, Kerberos, LDAP, Ranger, Knox and encryption in transit.

Why the split architecture matters

The blueprint does not place every workload on the same substrate. It keeps the stateful Cloudera Base layer in VMs while Data Services runs directly on bare-metal OpenShift workers. That gives organizations a migration path for existing VM-oriented data infrastructure without requiring an immediate container rewrite, while preserving a Kubernetes operating model for the newer services.

The approach also makes its overhead visible. Red Hat recommends reserving roughly 5% to 10% of host memory for the hypervisor and sizing storage controllers for concurrent I/O before moving stateful workloads. Those constraints are more useful to an architecture review than the broader consolidation pitch: they identify where density and performance assumptions need local testing.

What platform teams should do

Treat the published topology as a sizing baseline rather than a drop-in bill of materials. First verify the currently supported OpenShift, RHEL and Cloudera combinations. Then reproduce storage throughput, memory-reserve and node-placement assumptions with the organization’s own data profile. Teams should also decide whether the operational benefit of hosting Cloudera Base in OpenShift-managed VMs outweighs the cost of maintaining two OpenShift clusters in this reference design.

The important development is not merely that the products can coexist. Red Hat and Cloudera have exposed the node counts, component versions and functional tests behind the claim, giving infrastructure teams something measurable to challenge in a proof of concept.