An airport operations data center wants shared high-throughput model checkpoints across many trainers and is evaluating Fibre Channel SAN. When is that choice appropriate?

An airport operations data center that needs shared high-throughput model checkpoints across many trainers can appropriately evaluate Fibre Channel SAN when the design requires enterprise block performance, multipathing, and predictable shared volumes. FC SAN remains a valid AI storage option in Domain 2.3 whenever those enterprise SAN characteristics match the checkpoint sharing pattern—not because 'AI always means object,' but because the workload needs shared, multipathed, high-throughput block access. Expecting enterprise SAN/FC with object versioning fails by service model: FC SAN provides shared block volumes, not object storage with versioning, so versioning does not describe FC SAN for shared checkpoints. Requiring enterprise SAN/FC over native IP networks fails by transport: FC SAN runs on dedicated Fibre Channel fabric, not native IP, so native IP describes iSCSI or NVMe/TCP alternatives, not FC SAN for shared checkpoints. Saying enterprise SAN/FC with single-path connectivity is appropriate defeats the stem's requirement: single-path removes multipathing for failover and aggregated throughput, leaving a single point of failure for concurrent checkpoint writers. Exam caveat: Appropriateness is conditional. FC SAN is not mandatory for every AI lab, but blanket rejection is equally wrong. Match protocol to the shared high-throughput checkpoint use case described in the stem. Operational check: size FC fan-in, zoning, and multipathing for concurrent checkpoint writers; validate throughput and failover with representative trainer counts; compare against Ethernet/NVMe-oF alternatives using the same airport checkpoint SLA rather than ideology about fabric type.