A port authority DC must isolate PHI inference traffic from public kiosk networks. What network evaluation focus applies?

A port authority DC that must isolate PHI inference traffic from public kiosk networks should focus network evaluation on security segmentation that keeps sensitive AI tenants separated from public-facing traffic. VRFs, VLANs, and policy boundaries keep PHI inference away from lower-trust kiosk domains so a compromised public endpoint cannot freely reach sensitive AI workloads or their data paths. That separation is a first-class fabric requirement alongside bandwidth for regulated municipal AI.

QoS evaluation that prioritizes inference bandwidth guarantees for latency-sensitive tenants ensures performance for sensitive workloads but creates no trust boundary with public kiosk networks; guaranteeing bandwidth does nothing to keep PHI inference separated from lower-trust traffic. Throughput analysis that validates east-west capacity for distributed training replication confirms fabric capacity for replication flows but does not isolate PHI inference from public-facing kiosk domains. Latency optimization that minimizes hop count for inference serving across spine-leaf fabric speeds inference serving paths but provides no segmentation to keep regulated PHI traffic away from kiosk networks.

Exam caveat: Security segmentation of AI tenants is scored alongside bandwidth and latency when evaluating AI network deployment—especially where PHI or other regulated data is involved. Operational check: verify PHI inference and kiosk traffic never share a flat domain or a path chosen for throughput or latency alone, confirm management interfaces are not Internet-exposed, and test ACL/VRF policy with a denied cross-tenant probe before go-live.