Many corporations have a model-driven approach to their data nowadays. Any change is first modelled in a data-modelling tool like PowerDesigner or ERwin and only after it has been well thought through, with all possible impacts analysed, is it implemented in a DWH or other data system. Such an approach can be slower but in the long run it is very beneficial for the company, because to model each intended change means to document it before it happens. This way the documentation of data becomes an integral part of the analytical work, not a last-minute project phase, which is often undervalued or skipped.
Great information but for so few people
Bottom line: the model driven approach to data prevents chaos in a DWH and sustains its credibility and usability. It makes the data more understandable.
But here's the rub - a company with data models has a lot of information about data and its various flows - information which is, potentially, very useful to many users. In reality... it's used only by a few.
Why? Modelling tools like PowerDesigner or ERwin are focused on the authoring community, the technicians and analysts who know how to use them. They're not set up to share information across a company and let this information flow freely.
Giving away Jewels
The information in modelling tools like PowerDesigner or CA ERwin is potentially very useful to many users. These people don't need to be able to modify or create anything, they just need to be able to read the information and study diagrams related to it. They might also need to make searches, ask questions and comment on what they're reading - but that's about it. Which is where the idea of modelling-tool connectors comes from.
A modelling-tool connector is an application which loads important information from a modelling tool (texts and diagrams) and creates a web presentation from it.
Example: Imagine that each PowerDesigner model has its own web page. The page contains a list of all its tables, views and other objects. It shows all the diagrams related to the model as clickable pictures, ie. when you click on an object in the diagram, say a table, you're navigated to the page that describes the table in detail (setting out what columns, indexes and other metadata it has).
The effect - jewels: the modelling-tool connector gives away free and easy access to all the powerful and expensive data documentation, prepared by and previously accessible to a handful of analysts. Any user with a web browser has just gotten rich.
Semanta has a PowerDesigner connector and is currently working on an ERwin connector. Both of them load information extracted from the modelling tool, into a knowledge management system with a powerful full-text search engine. Thanks to this, any user can search through all the models, tables and other objects documented by the modelling tool, instantly.
This is even beneficial for users who have the modelling tool installed on their PC. After all, it's a lengthy and difficult process to find information in PowerDesigner and ERwin when you don’t know which model it's kept in.
The full-text search function of the knowledge system lists all the objects containing the text you've entered very clearly. If you need to look deeper, you can refine your results with preset filters - “show me only tables containing my text”, “show me only tables and only such tables which are in such models” etc.
Another nice side-effect is the social aspect: the modelling connector loads information into a web-based knowledge system which, among other things, allows user discussions on each of its pages. A user can read information from a model in his web browser and immediately ask the data-analyst responsible for anything he needs - not by email, but directly on the page so other users can also benefit.
Reality Check - allied connectors
“First model the change - then implement it” - a nice rule but one that it's not always possible to keep. Wouldn't it be interesting to be able to compare what's in your models and what is really in the DWH and other data systems.
This is something which can be achieved easily with an alliance of connectors. E.g. you have a Teradata DWH, you model in PowerDesigner, so you use a Teradata connector and PowerDesigner connector. Both connectors load information about the data-structures into one knowledge management system.
Thanks to the full-text search engine you can easily navigate from the model of a data object to its real metadata in the DWH.
And you can easily make a web page listing all the differences between the model and the DB reality.
Living in an imperfect world
Not all data-structures are completely modelled (this is even more true for data-flows). Often the DWH target is modelled but the stage or source systems are not. However, the combination of modelling and database connectors that I mentioned above, can create a powerful knowledge system even in a “partly-modelled environment”. This way you can get an overview of and insight into what is the true state of your data, even when your data-governance is still imperfect.
Our experience is that sharing data models with a broader audience, having it easily available and open to feedback leads to both a better understanding of the information and boosts its re-use, reducing communication overheads. Many companies have understood this and given the lack of a reasonably-priced software solution already on the market, have tried to develop a data-model sharing system on their own. They've found it both expensive and time consuming. Today, it's our pleasure to introduce to you an intelligent product that will fill this gap and will do it for a fraction of the time and cost.
You're welcome to check it out and try it at our UNLOCK your PowerDesigner page.