Tuesday, November 20, 2012

Colors of The Data Vault

Introduction

Most Data Vault modelers color code their Data vault models. Alas, the chosen colors usually differ between (groups of) modelers. Hans Hultgren was one of the first to start color coding practice (see http://www.youtube.com/watch?v=kRoDRlj8_YU ), but others like Kasper de Graaf did also use these color coding as well.

The first practice to my knowledge was the following coding:
  • Hubs: Blue
  • Links: Red
  • Satellites: Yellow/Orange or Yellow/Green or Green
  • Reference Entities: None, but I use an unobtrusive grey/purple
I also use a shape coding as well:
  • Hubs: square/cube
  • Links: Oblong/oval
  • Sats: Rectangle/flat
  • Reference entity: circle
  • Hierarchical Link: pyramid
  • Same-As/From To link: Arrow like construct

Using Color Coding Practices

Color Coding can not only be done on the basic DV entities, but also on attributes, relationships and keys and hence on any style (Normalized/Dimensional) and other modeling technique (FOM's like FCO-IM). They provide an interesting visualization of modeling and transformation strategies. One of the best diagramming techniques to use color coding are FOM's like FCO-IM because it let's you track your transformation rules and you only need to color code roles and entities (whereas in ER you will be color coding keys, attributes and relationships)

Color Coding Data Models

Color coding data models is i fact a transformation strategy visualization method. It should rely on classification metadata of the underlying data model.
I'll show here some examples of color coding data models.

FCO-IM


In FCO-IM you color Nominalized fact types (hubs, links and cross reference tables) or all the roles of the fact type. You can aslo color code UC's role connectors as well

Data Vault

Here we usually just color code the basic entities.

DV Skeleton

The Dv skeleton is just the DV without the satellites.


3NF

In 3NF each and every entity can produce a hub, link and sat.


DV source system analysis

Here I show part of a detailed color coding of a 3NF source system that needs to be transformed to a Data Vault. See that not only entities, but also keys, attributes and relationships have extensive color coding as well as corresponding classification metadata.

Dimensional Model

A dimensional model shows a link as the basis for a fact table, with measures coming from a satellite. Dimensions are usually hubs with their accompanying satellites.

I hope this short overview will show you how you can use color coding the enhance the usefulness of your data (vault) model diagrams.