A Microsoft Fabric workspace contains scheduled Dataflows Gen2 that refresh several Lakehouse tables overnight. The team reports that some refreshes intermittently fail with throttling errors during the busy window, and they suspect the workspace lacks enough compute. Before tuning each dataflow, which workspace-level setting most directly determines whether these Dataflows Gen2 refreshes have the compute they need?
- AThe capacity assigned to the workspace, because Dataflows Gen2 refreshes consume compute from that capacity and an undersized or saturated capacity throttles them. Correct
- BThe workspace Spark environment default, because Dataflows Gen2 execute on the same Spark pool as notebooks and inherit that pool's node count.
- CThe OneLake workspace setting that controls the storage region, because refresh throttling is caused by the distance between the dataflow and the destination tables.
- DThe domain assigned to the workspace, because the domain sets a refresh concurrency ceiling that the overnight dataflows are exceeding during the busy window.
Why A is correct: Dataflows Gen2 run on the Fabric capacity backing the workspace, so the assigned capacity and its available compute units govern whether concurrent overnight refreshes are throttled, making the capacity the most direct workspace-level lever.
Why B is wrong: It is tempting to assume all compute is Spark, but Dataflows Gen2 use the dataflow engine on the capacity, not the notebook Spark pool, so changing the Spark environment does not address dataflow refresh throttling.
Why C is wrong: Region affects latency in some scenarios but throttling errors signal a compute limit, not a storage-region distance; adjusting OneLake region settings would not relieve the capacity pressure causing the failures.
Why D is wrong: Domains group workspaces for governance and do not allocate compute or set refresh concurrency limits, so changing the domain assignment has no effect on whether the dataflow refreshes have enough capacity.