Your organization is designing a data lake architecture on Google Cloud to support various data ingestion pipelines that collect unstructured data from multiple sources. After the data is ingested, it will be used in several processing pipelines to build a recommendation engine for website users. The structure of the incoming data may change at any time , and it is crucial to store the data exactly as received so it can be reprocessed later if the current pipelines fail to handle the structural changes. What is the best architectural approach for storing the ingested data to meet this requirement?