I have an interesting challenge for my readers today. You have fifteen minutes to explain a data model to a non-data architect. You must choose five key points about data models and explain each of them in three minutes or less. Ready. Set. Go.
Tom’s 5 in 15
- Why model data – Would you set out on your vacation to an unexplored area without your GPS or map? No. You need some direction when venturing into the unfamiliar. Have you ever been given only written directions without a map? One wrong term can quickly disorient your journey and have you searching to get yourself back on track. The data model is your map to guide and save you time in your data journey. It shows major points of interest and paths between those points of interest. It gives a picture of your data and helps you visualize how it fits together.
- What are the parts of a data model – There are boxes that are called entities. Data is placed into boxes that represent people, places and things; such as customer, vendor, order, and address. Each of these boxes contains the things we know about the person, place or thing called attributes; such as name, date of birth, sex, and race. Lines called relationships connect entities and pull related data together. Some examples are a relationship that relates customer to their home address or another that relates the customer to the orders they placed.
- How to read a data model – A data model is a visual presentation of sentences that form the business rules of the data. The entity boxes are the subjects and objects. The text on the relationship line is the verb. The relationship line contains notation on the ends: 0 translates to zero, | translates to one and < translates to many. They appear in pairs. 0| translates to zero or one, || translates to one and only one and 0< translates to zero or many. Start your sentence with the 0| or ||. The relationship between customer and order ( || places 0< ) reads “One customer places 1 or many orders.” Move through the model by forming sentences from each of the relationships.
- Why relationships are important – Let’s go back to my comparison of a data model to a map. What would you do if I handed you a map and told you to show me how to get from point A to point B? You would highlight the roads that will get you from the origin to destination with the turns along the way. That is exactly why relationships are important. They send you on the right path through the data. These relationships enforce rules that assure the data is reliable and returns the correct answer. Understanding relationships is important to assure that you get the answer you are expecting.
- Why metadata is important – The entities, attributes and relationships pull together the big picture of data. A data model is more than a diagram. The data architect has documented the model objects in what is known as metadata. Metadata details the characteristics of the data objects. Some examples of metadata include descriptions, field size and examples. Metadata gives context to the data and how it is stored and used within the organization. The data architect can give you a data dictionary of the objects in the data model.
There actually is a purpose to this exercise. As data becomes more accessible to people across the organization, there is a need to explain this data and how it relates to other pieces of data. The data model is one of the primary means by which this conversation occurs.
Many new data consumers have a limited amount of time to orient themselves to data models. They come knocking on the data architect’s door to understand the wealth of data they now have in their hands. The data architect must succinctly explain data models in 15 minutes or less. It will not be the only conversations you have about data models. It will be the most important one since it engages them in the use of the data model.