Hacker News Reader

Comment by maxerickson | Hacker News Reader

I continue not to understand the "reverse" in that name.

Like if you aren't already doing "transform", you probably aren't reversing it.

Traditional ETL tools focused on getting data out of services and into a few places like a relational DB, S3, or some other such aggregated storage so they were read-only from services and read/write to DBs.

Reverse ETL is marketing to differentiate to the customer that you're pushing data into services as well, so that it's read/write to all.

Of course the concept is amusing to someone talking about ETL, but the market moved at some point to refer to the first functionality as ETL and so this is just a change to differentiate.

Parent didn't chose the best starting place to illustrate.

Typically, reverse ETL starts with "all of my data is already in an EDW" and goes to "I want to push it to my various app systems."

Canonical example: pushing system of record information to Salesforce to speed and enrich the sales cycle.

https://hightouch.io/blog/reverse-etl/

My bad. My introduction to the topic was through the lens of different use cases

Np! Wasn't intended as a slight. We all come into new things the way we do, rarely via the textbook definition (which typically doesn't even exist when we're first doing it!).

Still, my understanding is that ETL predates the whole "data warehouse" thing.

Like "extract" isn't any more specific than a source and "load" isn't anything more specific than a destination.

I’ll concede that the E, the T, and the L are the only operations I see in this model

Reply on news.ycombinator.com