FREE email newsletters 
The Federated Future
Data Warehouse Delivery
Information Management Magazine, January 2000
Last month we saw how market and business forces were moving inexorably toward a fate of multiple, non-integrated business intelligence (BI) systems, rather than the "ideal" monolithic, hub-and-spoke enterprise data warehouse favored by various vested-interest vendors, theorists and idealists. Indeed, rather than the clean, pure world promised by the data warehouse revolution, a typical organization has evolved into a non-integrated architecture similar to that shown in Figure 1.
Advertisement
Figure 1: Non-Integrated Architecture
As we saw last month, the market and business forces that have led us to this place are not going to go away. In contrast, they will be increasing, and no IT-led argument about the elegance or potential dollar savings of a long-term investment in the idealist vision of a single hub-and-spoke enterprise architecture is going to slow them down. As those of us who live in the field know, the only word in the previous sentence the business will hear is "long," which doesn't square with the quarter-to-quarter performance metrics they live under (otherwise known as the sword of Damocles). Ergo, we witness the ongoing proliferation of non-architected data marts and non-integrated, proprietary "favored son" ERP data warehouse/analytical application systems.
To survive in the current and future world of low-cost, turnkey analytical applications and proprietary ERP data warehouse systems (it's architected, but whose architecture is it?), you must transition to a federated architecture as illustrated in Figure 2.
Figure 2: Federated Architecture
A federated environment is viable and sustainable for the following reasons:
- It doesn't try to swim against the tsunami of market and business forces driving non-architected/non-integrated systems. Instead, it facilitates the integration of these systems, thereby avoiding the certain political death of opposing a powerful executive's tactical agenda.
- It enables the goal of a single version of the truth. By sharing key metrics, measures and dimensions (masters) across the entire range of BI systems, it allows each system to manifest a consistent reflection of key business indicators and measures while also allowing each system to maintain proprietary measures within its captive environment. An example would be corporate net sales vs. marketing net sales vs. finance net sales.
- It increases BI flexibility. By creating an architecture that can accommodate the shifting needs and priorities of the business, the BI team is viewed as a political positive and key players in responding to quick changes in competitive or regulatory contexts. Rather than being an obstructionist team hiding behind the shield of a one-size-fits-all, hub-and-spoke architecture, the team can quickly and successfully accommodate custom, turnkey and proprietary systems.
Just as when the Model T appeared on streets previously occupied only by the haughty, moneyed classes in their custom-made motor coaches, you should expect much jeering, expletives, spittle and other forms of derision (chief among them the time-honored fear, uncertainty and doubt) from the ivory-tower crowd related to the federated approach. In their isolated, insular world, devoid of the cut and thrust of the Machiavellian world of business politics, the hub-and-spoke shines eternal in its shimmering light of perpetual elegance. For the rest of us in the trenches tasked with making these things work and delivering value to the enterprise within our political lifetimes, the federated architecture is our best route to survival.
Douglas Hackney is the president of Enterprise Group Ltd., a consulting and knowledge-transfer company specializing in designing and implementing data warehouses and associated information delivery systems. He can be reached at www.egltd.com.
For more information on related topics, visit the following channels:
