Use the starter deployment (formerly "demo") of CP4BA: the operator installs everything with a built-in database and LDAP, generated secrets and minimal sizing, in one namespace, through the cp4a-deployment.sh script or the OpenShift console form. For workflow choose the pattern Workflow authoring (optionally with case and BAI).
# starter deployment (interactive script from the case bundle of the release)
./cp4a-clusteradmin-setup.sh # cluster prerequisites as cluster-admin (storage classes, operator install)
./cp4a-deployment.sh # choose: Starter, patterns: Workflow Authoring, optional components: case, bai
# result: one namespace with operator + CR "icp4adeploy", built-in PostgreSQL, LDAP (with sample users), Studio, Workflow Center + a Workflow Server
oc get icp4acluster icp4adeploy -o jsonpath='{.status.endpoints}'Practical numbers: a starter with workflow authoring + case fits in roughly 24-32 vCPU / 64-96 GB RAM worker capacity and 100-200 GB of storage (the exact minimum is in the release's sizing table); BAI adds a lot (Kafka, Flink, OpenSearch) - leave it out for developer sandboxes. Limits of the starter profile: not supported for production, built-in database and LDAP are not backed up by IBM procedures, single replicas, self-signed certificates; upgrades are possible but reinstalling is often simpler. Alternatives: one shared authoring environment per team (Workflow Center is multi-user by design - developers share it like a Process Center) with personal Workflow Server namespaces only if playback isolation is needed, or Workflow Process Service as a container (even smaller) when case is not required. IBM also offers time-boxed CP4BA trial environments (IBM Cloud / TechZone) for sandboxes without your own cluster.
References