The core philosophy of a data lake is "never touch the original data, and accumulate refined copies stage by stage." Today we cover S3 zone structures, partitioning, and naming conventions when using Amazon S3 as data lake storage.
S3 provides virtually unlimited capacity, 11 9's (99.999999999%) durability, and separation of compute and storage. This separation allows multiple engines like Athena, EMR, Redshift Spectrum, and Glue to read the same data simultaneously. Unlike traditional data warehouses, data lakes fill storage first and apply schema at read time (schema-on-read).
💡 Related Theory: Data lakes use schema-on-read, while data warehouses use schema-on-write