Question

A data engineering team is designing a medallion architecture for streaming IoT sensor data. The raw events arrive as semi-structured JSON with inconsistent schemas and occasional duplicate records. The team wants the first landing layer to preserve the data exactly as received from the source, including any malformed rows, so they can reprocess if downstream transformation logic changes. In which layer should this raw data be stored, and why?

Accepted Answer

The bronze layer, because it captures raw ingested data as-is to preserve full source fidelity and enable reprocessing

Answer

The silver layer, because it is the entry point where all incoming data is first cleaned and deduplicated

Answer

The gold layer, because raw data should be aggregated immediately to reduce storage costs

Answer

A temporary view, because raw data should never be persisted to Delta tables