Data lake ingests the operational systems’ data in its raw form.
Eventually, when raw data cannot fit the needs of data analyst, they can create their ETL scripts to generate analytical models and save them in the data warehouse.
Because of the fact that we store original raw data and only transform it afterward when needed, we can create multiple analytical models for different use cases.
Data lake are schema-less. If there is no control over the quality of the incoming data, data lake can become data swamp.