In this blog article we are going into the gears and bolts of the dashboard; the back-end supporting it. Here are some tips regarding the infrastructure of dashboard implementation.
A dashboard implementation needs to be fully integrated, with the data model underneath to provide adequate response time. This means that the data modelers should know the functional data requirements of the dashboard, as early as possible.
Dedicated data model
Sometimes a dashboard's data layer requirements are not compatible with the enterprise data warehouse. In these situations it is wise to invest in providing it with its own dedicated datamart. At agileDSS we have developed a proprietary generic data model for such situations successfully used with some of our clients.
Bridging the gap to the ad-hoc
The dashboard user might want to access the underlying data, to query it in other axis of analysis or some other analysis need. The presence of conformed dimensions to provide quick drill across, is highly useful in these cases.
Push the logic down in the semantic layers
Generally speaking, when dealing with BusinessObject's universes we are all tempted to enhance the universe layer with business logic. This is a good approach provided the universe is a shared layer with other BI applications and there is a single consumer platform technology (Business Objects)
The rationale being to avoid duplicating the business rules in other medium. If the universe layer is the unique layer exposing the metric and that all consumer applications can connect to it, then adding to the universe is justified. This is a core principle of the BusinessObjects BI Model.
The other scenario would require letting the universe expose the metrics by pushing the business logic down in the data warehouse. Simple filters and other user-friendly components could then be added to the universe, as long as they do not affect the integrity of the measures.
Tips From The Trenches: QaaWS Connectivity
Business Object's Query as a Web Service is a method for Xcelsius to connect to Universes. Since it is based on a web service communication standard we tried integrating it with a Java based Dashboard. We finally made it work, but the integration was not a smooth road and added time was required. I would therefore warn against doing this implementation.
I hope these tips will help you build a stronger and tighter dashboard integration!