Interview With Wayne Eckerson, the Data Warehouse Institute
Wayne Eckerson is director of research and services for the Data Warehouse Institute (TDWI), an organization that provides education, training, certification and market research for business intelligence and data warehousing professionals worldwide. Information Builders recently spoke with Mr. Eckerson to learn about key issues facing today's BI professionals. Some of these concepts are explained in greater detail in his new book, Performance Dashboards: Measuring, Monitoring, and Managing Your Business (Wiley, 2005).
IB MAGAZINE: Our customers are increasingly interested in operational BI systems and operational dashboards. How do you define this branch of business intelligence?
ECKERSON: In many ways, operational BI doesn't stand for anything new. An operational dashboard is merely the window through which workers, managers, and executives can monitor business processes and take action to avert a problem or capitalize on an opportunity. What's new is the immediacy with which they can obtain information. Unlike many traditional data warehouse systems, which might be updated once a day or once a week, operational dashboards let you analyze events closer to the present time – not necessarily in real-time, but generally with very little latency. Many users call this "right time information delivery" or "right time data warehousing," which simply means receiving information in time to positively affect an outcome. That could mean every minute, every hour, every day or even every week, but generally it means intra-day or less. Once you see an event or condition taking place, an operational dashboard should allow you to drill down to determine the cause, intervene in the corresponding process, or perhaps generate an operational report to investigate further.
IB MAGAZINE: Do operational dashboards supplement other types of reporting and BI systems, or replace them?
ECKERSON: As I describe in my new book, Performance Dashboards, what most users really want is a layer on top of their analysis, reporting, and planning tools that pulls these disciplines together via a highly simplified and intuitive interface. This is what we think of as a dashboard or scorecard, since it enables business users to track and monitor the metrics they care about most. It also helps them compare actual performance to predefined targets, and it triggers alerts when performance strays too far from goals.
Unfortunately, most companies build different applications to handle different analytical tasks, such as real-time monitoring, slice-and-dice analysis, enterprise reporting, and scorecarding. These analytical silos are costly and redundant and make it impossible for a company to ensure that everyone uses the same metrics and data.
IB MAGAZINE: What type of organization is typically interested in an operational BI system or operational dashboard?
ECKERSON: Just about every organization has a need for right-time information. The question is, is the value of having timely information worth the cost of delivering it? In my book I use the example of Quicken Loans, the nation's largest online lender, which closed $12 billion in retail mortgage loans in 2004. Any outage or slow-down in this core operation can cost Quicken Loans millions of dollars per hour. To stay abreast of what is happening on the sales floor from one moment to the next, they implemented a series of operational dashboards.
IB MAGAZINE: What are some of the technologies you need to build these operational systems?
ECKERSON: Operational dashboards are generally part of a BI infrastructure that merges operational and analytical processing into a seamless whole. Although operational dashboards can be constructed in many different ways, there are a few indispensable technologies that make them possible, such as enterprise application integration (EAI), enterprise information integration (EII), active data warehousing, service oriented architecture (SOA), and operational data stores (ODS).
IB MAGAZINE: How do you build these right-time BI environments.
ECKERSON: The best way to build a right-time BI environment is to trickle-feed data into it via an event-driven messaging backbone. EAI software is a good way to extract data from source systems in real time. For example, Information Builders customers can use the iWay technology to create a copy of each event or transaction as it occurs in the source system and publish it to a messaging backbone. Then, any application on the backbone can subscribe to the event or message, grab it off the backbone, or store it in a real-time data store. Quicken Loans developed a Web service to capture events from the messaging backbone and store them in a real-time data store, which is accessible through the dashboards.
IB MAGAZINE: Are real time analytics, event monitoring, and business activity monitoring important aspects of operational BI?
ECKERSON: There is a lot of overlap, and in some cases these terms refer to the same thing. Event monitoring generally involves waiting for a specific thing to happen—perhaps by polling a database for a given condition or relationship—then triggering an alert or a process when that condition occurs. Analytics is generally what happens after you get an alert. It might prompt you to analyze a situation once an event occurs, and a good dashboard will help you do that. Business Activity Monitoring is similar but more comprehensive: it applies this type of strategy to an entire value chain. All of these things fall into the category of operational BI.
IB MAGAZINE: What are some of the cultural issues that go along with this type of operational analysis?
ECKERSON: There is no point in developing a right-time BI infrastructure if companies don't reengineer core business processes and systems to exploit the information. For example, why provide store managers with hourly sales data if they can only change prices or modify shelf displays once a day? If a company thinks it is important to deliver hourly information to managers, then they must upgrade the corresponding systems so managers can exploit the information.
There are also basic issues of usability. Ultimately, the purpose of an operational dashboard is to empower users to work more proactively and make faster, smarter decisions. However, the human aspects of these projects are often the least reliable component. You must train workers to use the system proficiently and make sure they know how to interpret the data they are seeing.
IB MAGAZINE: For a while the analyst community was talking about enterprise business intelligence suites (EBIS). Now we're hearing about BI Platforms. What's the difference?
ECKERSON: BI vendors have been talking about EBIS for ten years. It began with a trend to provide many types of tools within one BI family – OLAP, query, parameterized reporting, production reporting, multidimensional analysis, and so forth. These suites got bigger and bigger, but many of them lacked integration, meaning the various components still behaved like individual products. A BI platform is much more tightly integrated. At the foundation is a common architecture upon which you layer BI functionality—a common application development environment, a common security layer, common meta data, a common repository, a single API, a single reporting engine, a single information delivery engine, and so forth. Each of these functions takes the form of services that you can access via a service-oriented architecture. We're not there yet, but this is where the industry is heading.
IB MAGAZINE: What are the advantages of this type of platform?
ECKERSON: It's easier to install, easier to maintain, and more cohesive all around. An organization can master one platform and have all these services available to them. Instead of buying all these distinct products separately, you use modules within the platform. All of the BI tools are integrated at the architectural layer instead of at the user interface or front-end layer. Again, the difference between a suite and a platform is the degree of integration among the toolset. Are you providing a set of tools or are you providing a set of services running on a common BI architecture? That's a question for your BI vendor.
IB MAGAZINE: How important is the data architecture in these BI platforms?
ECKERSON: It's fundamental. A BI architecture needs to run on a data architecture. BI platforms are comprised of well-integrated tools or modules that are aligned with business strategy and run on a common BI and data infrastructure. This allows you to deliver a single version of the truth whether your BI applications run on a data architecture implemented using a data warehouse or EAI or EII technologies or something else. A BI platform needs a comprehensive integration environment to make this possible. Information Builders' WebFOCUS uses iWay technology to implement such a data architecture.
 
|
