A team proposes ingesting only the aggregated totals their current dashboard shows, rather than the underlying detail. What risk does this create?
Select an answer to reveal the explanation.
Short Explanation
Aggregates answer today's question and nothing else. The first time someone asks why a number moved, the rows that would have explained it were never loaded. You need the detail in the lakehouse to reconcile, re-derive, and answer new questions.
Full Explanation
The correct answer is the one stating that any question that was not anticipated cannot be answered, and figures cannot be reconciled or re-derived because the detail no longer exists in the lakehouse. Aggregation is lossy. Loading only totals hard-codes today's questions into the pipeline, so a new breakdown, a reconciliation, or an investigation into an anomaly cannot be served without going back to the source - assuming it still retains the detail.
The option saying aggregated data cannot be stored in an open table format is incorrect because aggregates store perfectly well in a table format; the issue is what was discarded.
The option saying aggregates cannot be governed, since policies apply only to detailed records, is incorrect because governance applies to any registered asset regardless of granularity.
The option saying dashboards cannot read aggregated tables without the underlying detail present is incorrect because dashboards read aggregate tables happily - which is why the limitation goes unnoticed until someone asks a new question.
For the exam, ingest at a granularity that supports future questions, and aggregate downstream.