OpenShift hosted control planes (HCP) introduce confusing terminology that has evolved across upstream projects, product naming, and API conventions. This guide clarifies the key terms: management cluster (hosts the control planes), hosted cluster (the full logical cluster including workers), hosted control plane (just the control plane pods), NodePool, HostedCluster CR, and the distinction between managed cluster vs. management cluster. HCP decouples the control plane from the data plane, running it as pods on a management cluster while workers run on bare metal, VMs, or cloud instances. Benefits include lower infrastructure cost (no dedicated master nodes per cluster), faster provisioning (minutes vs. ~45 minutes), and centralized multi-version management. Trade-offs include management cluster becoming a critical single point of failure, new operational complexity, stricter etcd storage requirements (sub-10ms p99 WAL fsync latency), and no direct control plane node access. A decision framework helps determine when HCP is appropriate versus standalone OpenShift clusters.
