A pilot ran on a single-node local install and worked well. The team proposes the same topology for production, serving several hundred analysts. What should the review raise?
Select an answer to reveal the explanation.
Short Explanation
A pilot proves the thing works. It proves nothing about what happens when three hundred people use it at once, when one bad query monopolises the node, or when that node reboots. Those are exactly the properties production is judged on.
Full Explanation
The correct answer is the single-node topology choice, because a single node concentrates every concern into one failure domain. There is no headroom for concurrent load, no separation between a heavy batch job and interactive queries, and no continuity when the node is unavailable. A light pilot exercises none of these.
The claim that single-node installs cannot use object storage is incorrect because object storage is used regardless of topology. The storage layer is not what changes.
The claim that governance policies do not apply to single-node installs and would need to be re-created is incorrect because governance is a platform capability, not a function of node count. Policies apply in either case.
The claim that nothing should be raised because a working pilot topology is by definition validated for production is incorrect and states the fallacy the question is testing. A pilot validates function under pilot conditions; production adds scale, concurrency and availability requirements the pilot never touched.
For the exam, treat pilot topology carried unchanged into production as a design smell in almost any scenario.