When ELT emerged as the dominant pattern in data ingestion, it became clear that doing some transformations after extraction, but before loading, was still beneficial. However, instead of transformations involving business logic or data modeling, this type of transformation is more limited in scope.
Some examples of the type of transformation that fits into the E(t)LT subpattern include the ff.:
- Deduplicate the records in a table
- Mask or otherwise obfuscate sensitive data