Design Challenge #4 - Techniques in reviewing a data model
|
There are a number of techniques I apply when I review a data model. Using these techniques over the years helped me quickly summarize models and look for ways to improve them. Many times the 80/20 rule seems to prevail when reviewing a model. That is, in 20% of the time we can validate 80% of the model. (That extra 20% can be a bit tough however!) Many of these techniques I plan on one day publishing and making it a topic in my modeling course. An example of such as technique is: I look at each data element and quickly identify which data elements do not end in a valid classword. A classword represents a general domain for the data element and examples include AMOUNT, DATE, NAME, IDENTIFIER, INDICATOR, and CODE (there are others too). Finding data elements that don’t end in a valid classword represent a naming standard violation at a minimum, and can also signal a larger problem where multiple pieces of business information might be stored in the same data element. For example, is Gross Sales an amount, a weight, a value, or worse yet, a combination of these? That is, should the right data element name be Gross Sales Amount, Gross Sales Weight, etc.? Now I know there are a lot more tips out there that you might use when you review a model. Share these with me. I will compile some of these techniques (including more of my own) and send them back out in this design challenge response. Note that as with all design challenges if I use your technique (and it’s not already on my list) I will give you credit in any documentation. Looking forward to hearing your techniques! |