Data Integration and Management

Recently I ran into this article entitled: ELT vs. ETL: Much Ado about Something .

Yes. Yet another ETL vs. ELT blog.

I just couldn't help myself writing something. I would agree with the assessment that "There's no doubt that ELT is now a mainstream capability" ... now the only disagreement is on the how. I would disagree that the term push-down is the only mechanism for ELT; in fact there are multiple approaches.

Some solutions use a push-down approach, what I like to call "ETLT" (a pushdown of the transformation after the fact) while others natively run transforms on the target systems. There's certainly advantages to running optimizations directly on the target system natively without the overhead of having to do the conversions.

As data warehouses, business intelligence applications, yearn for real-time capabilities, the low latency requirements for these systems becomes paramount. This is especially true for data warehouse appliances that are designed to take full advantage of these types of native ELT approaches.

So I hope we are all in agreement that ELT is the right path. Send me your views if you agree or disagree on the how.