A data link must reach a switch across the hall, beyond what copper twinax comfortably spans, and the tech is weighing twinax against fiber. What should drive the media choice for this run?

Media choice for a link is a distance question answered against the support list: the measured run is matched to the platform's supported media, where copper twinax carries electrical reach limits engineered for top-of-rack and in-row runs, and runs beyond copper's range are served by optical assemblies with supported fiber and the correct optic type for the distance class. Selecting at the intersection of measured distance and documented support keeps the link inside engineering that has already been validated on the platform. Stretching twinax past spec fails by concept: copper reach is an electrical limit, not a suggestion, and a beyond-spec link that negotiates can still deliver error-ridden service — negotiation and clean operation are different claims. Trusting first-try link-up fails because association is the weakest possible evidence; margin shortfalls surface as CRC errors and flapping precisely when traffic load arrives. Defaulting every run to fiber fails economically and technically: every optic still has its own distance class and support requirement, and short in-rack runs gain nothing from fiber's connectors and cost while the selection duty does not disappear. Exam caveat: transceivers and cables must come from the platform's supported list regardless of media type. Operational check: record each run's measured distance, confirm the chosen assembly's rated reach and support status, and watch error counters after the link takes real traffic.