Where's Your Backbone?Effective enterprise data delivery requires more than just a traditional data warehouse
by Kevin Jahn When analyzing Monday morning exception reports, which have titles such as "Products Sold Without Contract" or "Outstanding Invoices With More Than 10 Items," it's common for many business users to ask themselves, "Why is this necessary?" or "How did this happen?" At that point, they open a query or online analytic processing (OLAP) tool and begin to run queries against tables that might have the answers, only to become confused by the hundreds of records involved. Simply put, we're overwhelmed by information. Where are the answers to our questions? What tables do we need to look in? What questions do we ask, and how do we ask them using these tools? Furthermore, how do we know if the data we're reviewing is correct? Usually, we ask the data analyst or someone familiar with the underlying data. Sometimes, we just make assumptions about what we're looking at in the report and move on. Data is, by nature, highly volatile. Some data sources are less volatile than others (such as data warehouses), but for the most part, data is constantly accumulating or changing. Whether it's in the Monday morning exception reports or late-afternoon operations reports, one thing is certain: Decision makers need access to the right data at the right time. Fortunately, a mechanism called enterprise data delivery can assist in getting the right data to the right data consumer at the right time through a common, consistent, and secure channel. Enterprise data delivery usually requires more than just a data warehouse or analytic infrastructure; it takes a lot of well-engineered resources working together in what I call a data cycling backbone. In this article, I'll explain the key components of this concept at an introductory level and offer a blueprint for data delivery solutions. You'll find that the required building blocks are complementary to the components of data warehousing, data marts, analytics, OLAP, and even enterprise application integration. (See Figure 1.) Understand Your Decision PointsIn an OLAP or analytic environment, a data cube might limit the amount of information users need to navigate. However, the reporting solution must match the specific business need. Various standard reports are used on a daily basis and information may be inconsistent among or within departments because of stovepiped database or reporting solutions. In truth, most users simply require standard reporting, and in many ways, preconfigured reports or query wizards can be more effective in delivering the information the user needs. Other users may require access to specialized analytic tools to measure and evaluate data.
Business performance and growth depends on the ability to transform transaction data into knowledge, build wisdom, and ultimately support enterprise planning, key decisions, and corporate strategy. This business intelligence (BI) cycle starts with understanding business decisioning goals and classifying decision points within the business processes. Key performance indicators are great for showing numbers based on critical business metrics, but they must be calculated in a manner that continuously keeps the business and chronological context in mind. That's because many complex business processes depend on users making expedient and informed decisions at specific points in time. Identifying these decision points is crucial. Traditionally, decision-making is supported by manually administered spreadsheets or standard reporting dependent on an IT team. It's important to take this idea a step further by defining and prioritizing decision processes that identify strategic, operational, and tactical decision points. These points become the basis by which you can build a functional data cycling environment and make reporting more consistent. Decision point identification can help introduce standards and rules associated with data and decisions. Decision rules acting as alerts and thresholds can help identify transient events or abnormalities that might occur in data and can also assist in making proactive business decisions. For example, if I were a manufacturing plant general manager, then I might want to be notified automatically when production capacity was down so that I could take appropriate action. However, in order to make that decision, I'd need data from my operational systems such as fleet production orders, current production capacity, sales forecast estimates, and even store-level demand data. A well-engineered data cycling backbone can serve both the rule-based automatic notification and data delivery from the operational systems. Yet, the basis on which we make our decisions is typically hidden deep within back-office, operational database systems: There's a wealth of information just waiting to be tapped. Where is all the data pertinent to my situation? Where did it come from? Who's responsible? Is it accurate? Can I get it now? What does it mean? As simple as these questions sound, we can't answer many of them with 100-percent accuracy. Data sources within an enterprise are available in many forms. Structured data typically comes from transaction or operational system sources, data warehouses, data marts, or OLAP cubes. Other systems might store unstructured data, such as documents, pictures, charts, blueprints, or spatial data with coordinates. The data that we're looking for is scattered across the company, which brings us to the second element of a data cycling backbone: metadata discovery. Calling All MetadataKey elements of information can be discovered across data sources, data processing routines, data warehouses and marts, cubes, and reporting tools. Many large businesses have departmental silos that have the same informational needs without being aware of their similarities. Data enters from many points between the original contact with the customer and the final transaction destination. This process can result in duplication of data and inconsistencies in business decisions. Thus, access to a consolidated view of metadata across the enterprise at any point in time is important for sustaining data consistency and organizational growth. A common component of an enterprise data warehouse or other types of reporting databases is a metadata management platform. Metadata describes critical elements of data scattered across the organization. Typically, it's difficult to keep metadata synchronized across disparate systems and ensure that it's up to date. Therefore, a combination of metadata discovery, configuration, and management is necessary to complete the cycle of metadata synchronization. As decision makers, we've all gone on information expeditions to find accurate and consistent answers to even the simplest of questions without knowing where to begin. Metadata integration can help manage and integrate the abundance of information directed across the enterprise and ensure the appropriate data is available to users. Delivering the GoodsThe decision-support environment is so complex with so many systems capturing data that it's very difficult to organize all the input into useful, integrated business information and ensure that data requests are serviced accurately and efficiently. Thus, the third and key element of the data cycling backbone is an enterprise data delivery engine that, in fact, has several jobs to do in parallel: Ensure secure data access, service requests from data consumers, integrate requested data sources, and deliver the data to one or multiple target devices.
|
 |
Most Popular This Week
IE Weekly Newsletter
Subscribe to the newsletter
| |
|
|
|