ETL IS DEAD
Bill Inmon
Founder, Chairman, CEO, Best-Selling Author, University of Denver & Scalefree Advisory Board Member
ETL IS DEAD
By W H Inmon
Recently I saw a post by Vinay Samuel proclaiming that ETL is dead. Vinay pointed to the telecommunications companies who have made a mess of their data warehouses and ETL.
I certainly have no direct knowledge of how the telecommunications companies managed to mess up their data warehouses and ETL.
But to proclaim ETL dead is simply wrong. It is like saying the rudder of the Titanic was to blame for sending the ship into an iceberg. The rudder was simply doing what it was told. How about putting the blame on the captain and the shipping line of the Titanic that wanted to make haste across the ocean. It wasn’t the rudders fault.
So what is wrong with the proposition about ETL that Vinay has outlined?
1)?????Nowhere in ETL does it say that you should place all of your transformed data into a single repository of data. In fact, the books on data architecture tell you to not do exactly that. ETL is merely for the transformation of data, not the particulars on where the data that has been transformed is stored. The consultant and/or the vendor that directed your organization in the architecture for the building of the data warehouse should have read beyond the buzz words in the brochure in order to understand what was being said. In a word, you can and should store the output from textual ETL in a variety of places, based primarily on the probability of access and usage of the data.
2)?????So who told the telecommunications companies to store the output from textual ETL in a single place? Does anybody remember the names of large mainframe firms and parallel computing firms that sold data warehouse and ETL? Or some consulting firms that purported to be experts in architecture? Most of the people in these firms read a few brochures, learned a few buzz words, then sold their services to the telecommunications companies. The vendors and the consultant never took the time to understand architecture and the many nuances of data architecture that were being discussed and that were needed by the telecommunications firms. The vendors and the consultant merely used ETL and data warehouse as a means to the sales of the latest product from the vendor. And the telecommunications companies bought what they were sold – hook, line and sinker. Now who is to blame here? The vendor? The consultant? The IT manager who looked no further than next Tuesday? They probably were all to blame. But the IT manager should have been smart enough to ask penetrating questions to determine the qualifications of the vendor or the consultant. But that would require the IT manager to take initiative and most IT managers don’t have an ounce of initiative. So there is plenty of blame to go around the room.
They were all to blame. It simply was not entirely IT’s fault that ETL was used inadvisably by the telecommunications firms.
3)?????Without ETL you can NEVER achieve an enterprise view of information. NEVER. Corporations cannot ask such questions as –
?????????????????How many customers across the enterprise do we have?
?????????????????How many sales across the enterprise did we make?
?????????????????How many customers did we lose across the enterprise this last quarter?
And unfortunately, these questions are at the heart of making business decisions. Corporations HAVE TO HAVE this kind of information.
Blaming ETL for the insufficiency of the larger data architecture that evolved in the large telecommunications firms is like blaming the sun for rising in the east and setting in the west.
ETL is a fact of life and is not going away anytime soon, if ever.
Blaming ETL just shows an ignorance of the facts.
Bill Inmon lives in Denver with his wife and his two Scotty dogs – Jeb and Lena. Jeb and Lena are out for a walk today because it is the first warm day of the springtime. They go by Mr. Larry’s house, and if Mr. Larry is out in the yard, Mr. Larry gives each of them a cookie.
Power BI Trainer and Consultant at BitanBit
9 个月Mladen Bare?i?
Director and Principal Tutor, Avancier Limited
1 年I've nothing against ETL, it is useful, and can help to decouple systems, but I am not sure that amounts to defence of it. Can't you a feed a data warehouse any way you like, via a messaging system for example?
Business Geek/Data Leader
1 年A decade ago people from a DW project I worked with started loading a data warehouse by using Java programs. They said "we dropped ETL because it was way too buggy". I opened the mic and asked if said Java program did read the data from the source system. Yes, said the tech leader. Also, I asked, does the program make any change in the data? Yes, of course, said him. Does it insert the data into another database, at the end? (scornfull he answered) Damn yeah you bet it! So you Java program that is NOT ETL just extracts, transform and load the data into another db. I am still waiting for his answer. (The project ended few weeks later, adopting Pentaho and having me training them.) People won't ever get tired of trying to scoup a little stage time by proclaiming Tech X is dead... affff...