BI Blog: Dashboard should be able to cope with multiple datasources

Enterprise Data Warehouses vs. Federated Data Stores

Creating a single enterprise data warehouse (EDW) has traditionally been the optimal solution to consolidating enterprise information. However, organizations have found that new data sources appear in the IT landscape faster than can be incorporated into the EDW. In contrast, some organizations have opted for a distributed architecture with information stored in several data marts and joined at the query level, often referred to as federated data. While both approaches have pros and cons, in either case BI software must effectively access multiple data sources to satisfy increasingly complex user requirements.

A true enterprise BI platform would be incomplete without the ability to access and query multiple information sources and present the information through a single standard interface. The user should be able to navigate across all five styles of BI—scorecards and dashboards, enterprise reports, OLAP analytics, predictive analytics, and alerting—irrespective of the data source. Then, business users can concentrate on solving business problems, rather than understanding data sources.

Dashboards must be interactive and highly visual to help users understand business performance. Dashboards that access multiple data sources typically fall into two categories. In the first case, a single dashboard can present data from many sources without joining data across those disparate sources. Overall enterprise KPIs can be monitored using this type of dashboard. In the second case, a single dashboard must join the data across heterogeneous data sources to perform the required analysis. For example, actual sales and forecast sales may be in two different systems. In this case, the BI layer must join the data across from the appropriate data sources. Of course, to perform joins between multiple data sources, common keys need to exist between heterogeneous data sources.