A HOUSE IN THE WOODS by W H Inmon

A HOUSE IN THE WOODS

By W H Inmon

It is summertime. It is hot in the city. You see a picture and an ad for a vacation house in the words. There is a lake. There is a canoe. There are trees. There is fresh air. What a nice place to escape to for a week or so. The picture looks really good.

You pack up your bags and head off to the countryside. When you arrive the house is just as it was portrayed. Rustic. Woody. By the lake. When you take your first steps into the vacation house, everything seems fine.

You have a bite for dinner and you settle in for the night.

The first thing you notice is that there are cobwebs. Then you notice the mice running around. Then the ticks come out. Then the spiders. But you are tired and you go to sleep listening to the night sounds of the forest.

The next morning there is no water for a shower or for brushing your teeth or for flushing the toilet. It turns out that is a good thing because the water – in the lake and everywhere – is full of e coli bugs.

Then you head for the lake and some morning fishing. The canoe has a hole in it and the paddle is broken. The fish aren’t biting because they are all dead. The water in the lake is rancid. This really wasn’t what you were expecting when you saw this beautiful picture of a vacation house in the woods.

This scenario is what happens when people have a baseline of expectations that aren’t being met by the advertisements.

When you first see the house in the woods, you expect that it will meet some basic expectations. You expect that it will be clean. That there will be running water. That there won’t be mice to run over your feet in the middle of the night. That the lake will support recreation. That you can flush the toilet. Imagine your surprise and disappointment when you find out that these basic expectations are not met.

Such is the scenario painted by the vendors that tell you that you don’t need ETL or ELT.

The vendor says just go out and map to your enterprise data and go get it. Don’t bother with that messy ETL/ELT stuff. ETL/ELT is just overhead that gets in the way. Who needs it anyway?

What these vendors don’t point out is that almost everybody’s enterprise data is full of nasty surprises. If you do just go get your enterprise data, be prepared for data that may give you indigestion. Just because the data is found somewhere in the corporation does not make it usable. Or correct. Or believable.  Do you believe data found on a spreadsheet? You shouldn’t. You have no idea how it got there or what it means.

Furthermore, enterprise data is notorious for being warped by its focus on an application. Enterprise data is made up of applications, and each application has its own way of shaping its data. In one application gender is “male/female”. In another application gender is “m/f”. In another application gender is “1/0”. Adding the data from different applications together does not make sense until the data is integrated. And the simple example of the lack of integration of data representing gender is merely the tip of the tip of the tip of the iceberg. Nearly ALL enterprise data is subject to the unintegrated influence of different applications on enterprise data. Merely jumping in and grabbing some enterprise data may lead you to some very incorrect and misleading conclusions. If you do just go out and directly map your enterprise data into your analysis, be prepared for the consequences of making incorrect and bad decisions in the corporation.

It is like drinking water from a stream in the forest. The water may be clear and pure. Or the very same water may be full of e coli just waiting to make you sick or even kill you. You can’t tell just be looking at the flow of water in the stream. You don’t know about the potability of the water unless you treat the water first.

So, if you don’t care about the quality and the believability of your data, then jump right in and do a direct mapping of your enterprise data without first transforming it. Go ahead and open that door of the vacation cottage. But be prepared to sleep with ticks and mice and spiders if you do.

Who needs ETL/ELT? The only people that need ETL/ELT are the ones that care about the accuracy and believability of their data. If you don’t care, then just jump in and have at your enterprise data.

Bill Inmon’s latest books are HEARING THE VOICE OF THE CUSTOMER, Technics Publications, TURNING TEXT INTO GOLD, Technics Publications, and DATA ARCHITECTURE: SECOND EDITION, Elsevier Publications. Bill’s company – Forest Rim Technology – reads raw text and turns that text into a standard data base. Bill lives in Denver, Colorado.