Question: 1/50
A solution architect is planning an IBM Cloud Pak for Data deployment on Red Hat OpenShift. The platform team requires that all workload pods can be scheduled even when one worker node fails, without changing application code. Which design choice best supports this requirement?
Deploy all Cloud Pak for Data services on a single worker node to avoid cross-node dependencies
Use multiple worker nodes and configure pod anti-affinity or topology spread so replicas are distributed across nodes
Disable Kubernetes health probes so pods are not restarted during node failures
Run Cloud Pak for Data services on control-plane nodes to increase availability