Moving Forward with the On Demand Real-time Enterprise

Chapter 3. Architectural considerations 75

There is a significant effort required to cleanse, reconcile, and integrate data

from the various data silos around the enterprise. However, to move to a

real-time enterprise environment, there is a requirement to do so.

Isolated decision making

In a traditional business environment, business planning or financial modeling is

typically an offline activity. Statisticians would gather and analyze huge volumes

of data from the enterprise data sources and generate a stack of reports

representing the results of their analyses. Then business analysts would interpret

those analyses and develop a business model and plan. The management

executives and officers then develop the budget based on data analysis and their

business intuition. As you can see, the activities are exclusively disparate and

disconnected. With any luck, the offline reporting and offline analysis might

provide companies with the expected value. Without luck, there is a good chance

that they will not be able to provide accurate and consistent information for

adequate business intelligence purposes.

Latency in the decision process

Most businesses would reap tremendous benefit if they could reduce the latency

between business events, information availability and access, and the

formulation of responses. For example, the slow sale of one product in one

region might be trivial and be overlooked. However, if it repetitively occurs in

many stores, the impact could be significant and require fast investigation.

Unfortunately, the business intelligence information needed to detect market shift

and customer attrition is too often not available to the decision makers in time to

initiate a proactive response. As a result, companies are more often reactive in

an attempt to minimize the impact of events rather than being able to avoid them.

3.1.1 The impact of data warehousing

Data warehousing has, over time, become the defacto primary information

environment for businesses. As an architected environment, it is flexible to

enable change, while at the same time able to deliver the capabilities needed to

support analytic applications, queries, and all forms of data access and analysis.

With the increased need of making data available faster, the data warehouse

environment now includes the operational data store (ODS). The challenge is to

architect such that the contents of the ODS and data warehouse can be

integrated to enable the required support for a real-time enterprise environment.

Operational and informational

The first and most fundamental paradigm of data warehousing is so obvious that

it is often overlooked. Data warehousing has always made the assumption that