Today I heard an interesting answer to the question : what is the difference between ETL and ELT ?
The answer is : in ETL the transformations are processed by the ETL tools while in ELT the transformations are processed by the target datasources (RDBMS). I guess that the rising scalability of databases allows this kind of data integration to be possible.
So the important thing to remember is the side where the transformations are processed.
I made the following schema to help remember the difference :
ELT is appropriate for databases like Teradata. I heard that the Teradata engine is extremely fast and efficient at ELT (Extract, Load, Transform) processing to aggregate and resave data.
ELT should be used, instead of ETL, in these various cases :
Talend Open Studio already integrates ELT components :