Blog: Dan E. Linstedt

« A funny idea: Slower Melting Snow | Main | Information Valuation - Part 2 »

Data to Information, Architectural Roles for Business

Dave Wells, Director of Education, TDWI and I have had several discussions on this topic: Turning your Data In to Business Information

In light of this discussion we discussed the Business Dimensions and Business points of pivoting which take place when layering the data for presentation. Data is often overlaid with additional business dimensions to make it usable. I'm not talking about the technical dimensions that we produce within the data marts, I'm talking about individual columns labeled as dimensional aspects of the data.

This isn't to say that parts of these ideas aren't available today; it is merely to say that some level of automation and underlying base data architecture are missing from the scene today.

For instance, there are the common and major dimensions: Sales, Finance, HR, Manufacturing, etc.. There are the other common dimensions such as: hours worked revenue, taxes paid, cost of goods, etc... But hidden within these are additional layers of business dimensions which we frequently ignore.

These dimensions are the most powerful - allowing the business user to slice and dice the data by column to reach a single cell of information. It's N-DIMENSIONAL information, something that could be utilized by data visualization engines. In this N-DIMENSIONAL space, we have all the other columns or data elements - but they are arranged in the manner in which we USE the data within business.

Wait a minute! Just hold on there partner, are you telling me that one of these dimensions could just be equated to a Satellite?
YEP! You got it, in the Data Vault modeling architecture; each Satellite could be created to become a (business) dimensional breakdown of the data itself. This might just be the business definition of a Satellite that we've all been waiting for.

Keep in mind that Satellites will still split by Type of Data and Rate of Change - this will help define the Business Dimensional aspect of the information housed within. Each column within the Satellite (at that point) becomes a pivot objective if desired.

So where's the challenge?
The challenge, in turning data into information, is in the nature of the utilization of this data and how it's organized for the business when presented. These multiple layers / independent stacks of data housed within satellites must be formulated to be extremely flat and wide, as if in an N-DIMENSIONAL CUBE, associated by snapshot in time. In other words, RDBMS engines with Cube-Views are ahead of the curve, BI vendors and middle-ware such as EII which can build cubes are also ahead of the curve.

The challenge is to connect EACH of these dimensional definitions in an X, Y, or Z axis for viewing when desired - allow the end-user to pivot on each of these dimensions, allow each of the dimensions to move in the hierarchy (up or down), and define them as full and complete metadata for the business users. In other words, a metadata repository for all elements in the Data Vault model, then make it accessible through cube-views or some such delivery mechanism.

The ultimate goal would be to have database technology powerful enough to collapse the business delivery into a virtually defined layer or layers, driven by metadata and virtual definition of the web of connectivity across the metadata. Then to have this layer be the single point of access by all BI and delivery mechanisms.

More about the Data Vault data modeling architecture can be found here, on free forum discussions.

Thoughts?
Dan L

Share: