Knowledge is power. An enterprise data warehouse (EDW) coupled with business intelligence (BI) is the means for providing the integrated analytic detail necessary to glean that knowledge within your organization. But how can you enable all of your employees to have access to that kind of information in real time? What could your organization do with that power?

Think of the hundreds of small decisions employees make each day. By having access to real-time data, all employees could make informed decisions, resulting in better service to customers and improved profitability. BI would pervade your operations.

This distribution of BI throughout the organization is a concept known as "pervasive BI." It's the key to a unified corporate vision and strategy. Companies are finding that pervasive BI enables better inbound and outbound cross-selling by customer service representatives, optimizes trailer and container loading and routing, enables dynamic pricing and helps detect point-of-sale fraud.

However, taking advantage of the many benefits of pervasive BI requires a different way of thinking about BI, not only from those who use it but also from those who enable it.

Improve information supply chain management
Since pervasive BI serves everyone in an organization, it is critical that the information supply chain seamlessly leverages the EDW. An infrastructure that is cobbled together will not suffice. The data infrastructure must be carefully planned and implemented to supply a flow of information that serves up "just in time" insights based on:

	Fresh, clean, accurate and complete data
	Integrated data from across the organization that is blended and highly available
	Three major sub-systems—data integration services, decision repositories and decision services

Delivering fresh data enables a company to remain agile and to react rapidly to situational changes. The "speed of need" determines an organization's freshness requirements and helps database architects design systems that deliver data at the correct intervals. But fresh data is not enough if it is not accurate. Using a data integration platform enables an organization to standardize, cleanse and match data across all of its data sources, allowing all of the relevant data to be linked and unduplicated in the EDW.

For data to be complete, it must be compiled from every possible source, including tapping real-time information sources such as message queues, enterprise service buses, Web services, extensible markup language (XML) interchange formats and other documents. Having complete data that is not highly available is meaningless. Keeping data highly available is a key requirement for pervasive BI. Downtime means lost business opportunities, so it is vital to minimize any planned or unplanned downtime for the EDW.

Data integration and decision services are the final pieces of the pervasive BI puzzle. Data integration services allow the bridging of multiple domains and can provide continuous streams of information in addition to batch file data acquisition. These services extract, discover, transform, cleanse and deliver data to multiple subscribers. Decision services, on the other hand, analyze the patterns, facts and relationships found in the enterprise's decision repositories (i.e., the EDW, data marts and operational data stores). Decision services also deliver the relevant information to the end users. These services include reporting, data mining, operational and tactical operations as well as strategic applications like market segmentation and risk analysis.

Serving diverse users
Once the necessary elements of pervasive BI are in place, how do you determine who gets what information and when? Serving a more diverse user population requires a wide range of analytical functionality and informational styles. The smooth handling of reporting, online analytical processing, ad hoc queries, continuous data loading, data mining, event-driven decisions and tactical decision support can be successfully accomplished only through mixed workload management and service level agreements (SLAs).

To manage workloads, companies can use proactive resource management such as query scheduling, which will determine when a query is sent to the database for execution. This will help alleviate over-committing resources and can smooth the way for completing each query in the set time frame. Another way to handle prioritization is through resource allocation as a query is pulled up for execution. For instance, if a tactical decision support query has a three-second SLA, that query will take precedence over a data mining query that can be completed in a two-hour window.

Clearly, greater expectations on the part of users—and the business—mean SLAs are a must. SLAs may cover processes and services that you might not have previously considered. In fact, a new set of service levels will be required to recognize the performance demands and end-to-end mission-critical availability requirements of an EDW.

Data SLAs, for example, may be required for:

	Real-time data collection. Gathers data from the origin systems and pushes it to the database
	Changed data capture. Enables the latest data to be captured and included in the database to support data freshness and completeness
	Micro-batch model. Utilizes small, frequent, scheduled pulls of information to gather such data as click-stream or call detail records
	Real-time resilience. Ensures the reliability of the data-integration services and underpins high availability

Other SLAs, such as high-availability metrics by the users, tactical query response times by user and analytics, and scalability in terms of concurrent users and delivery mechanisms, may also be advantageous.

Tie it all together
Like most business-critical operations, pervasive BI needs superior planning in order to produce excellent results. A slap-dash approach to setting up your data infrastructure will only cause frustration on the part of IT, employees and stakeholders. Careful consideration of your organization's data needs, the user community and your priorities will do much to facilitate a pervasive BI operation that yields satisfaction.

What could your company do with pervasive BI? Imagine how the flow of integrated data would enable faster responses, encourage effective operations, ease decision making and give rise to innovation in your company. Data can give all of the employees at your company the knowledge they need to do a better job and bring day-to-day activities in line with the company's strategic aims. Pervasive BI can make any company more powerful by giving its employees access to the right data. If you're interested in empowering your work force, pervasive BI is an effective way to accomplish that. T

Shirley S. Savage writes on technology subjects from her base in Maine.

Photo illustration by Randall Nelson

Teradata Magazine-June 2008