Continue in 2 seconds

A Taxonomy for BI

  • June 01 2004, 1:00am EDT

IT folks like buzzwords. To gain visibility and acceptance in the IT industry, new technologies and products must be associated with a buzzword - ideally a buzzword that the press and analysts are currently talking and writing about. If a new solution doesn't fit an existing buzzword category, a vendor may create a new word and then hope they can convince analyst organizations such as Gartner and META to adopt it. After all, analyst organizations make their money by creating new buzzwords and charging a fee to anyone who wants to understand it and anyone who wants to learn about products that are the best at supporting it. Vendors, in turn, view a good position on the vendor buzzword charts as crucial to success.

The above viewpoint may be cynical, but the fact is that we are a buzzword-driven industry. Buzzwords are useful, provided they are not abused. Even in today's business-driven approach to IT, where buying power is rightly in the hands of business users, IT must still install and maintain new technologies and products. Buzzwords enable IT staff to categorize new solutions, determine which department should be responsible for a new solution and select products which support that solution.

The problem is that once a buzzword becomes popular, everyone wants to be associated with it. This leads to buzzword abuse. Another problem is that the dividing line between certain technologies, and thus buzzwords, is becoming blurred. This is especially the case for data warehousing and business intelligence (BI). Business activity monitoring (BAM), performance management, dashboards, real time, process management and enterprise information integration (EII) are just a few examples of evolving BI technologies that are suffering from buzzword abuse.

When presenting on BI, I find that audiences seem to more easily understand a technology solution if they can associate it with a label or buzzword. One problem I face when talking about BI is not only ensuring my use of terminology is consistent, but also making sure that it is in line with the rest of the industry. Another issue for me is that often, not everyone agrees on a given definition for any given technology or buzzword. To help solve these problems, I have begun employing a BI taxonomy that relates BI buzzwords to business and technology needs. This makes it easier to understand and position any given buzzword. After all, we use taxonomies to categorize and understand business information, why not information technologies as well?

The first dimension to consider in a BI taxonomy (see Figure 1) concerns the BI applications that deliver information for business decision making and action taking. These applications support strategic planning, tactical analysis and operational action taking. These applications, in turn, consume and create different types of business information including business transaction data, key performance metrics, domain-specific metrics, actionable performance metrics (in-context scorecards) and automated alerts, actions and recommendations.

Each type of BI has a different currency or latency (real time, near real time, historic) and granularity (detailed, summarized). BI is also managed in different types of data stores (data warehouse, data mart, low-latency store, in-memory data cache, operational data store, multidimensional cube, materialized view, virtual view, etc.) and has different business rules associated with it (transformation, analysis or metrics, context, exception, action and recommendation).

BI applications can be off-the-shelf application packages, or custom-built in house. Application packages may be for front-, middle- or back-office operations and may address an industry-wide domain such as finance and/or a vertical industry domain such as telecommunications.

Vendors provide many solutions for building custom BI applications. These development solutions can be integrated into a BI delivery and collaboration suite, a BI development suite or a data integration suite. A BI development suite may contain a component-based interactive development environment (IDE), dashboard builder, business rules engine, business reporting, business analysis and business/corporate performance management capabilities. A data integration suite may contain data design and modeling, ETL (extract, transform and load), event monitoring, data profiling, data quality management, EII and meta data management facilities. In addition to the three suites of tools, a BI platform may include planning and prediction tools (data mining, planning, budgeting, forecasting, etc.).

Although this taxonomy is not exhaustive (it is constantly evolving), it should allow any given technology, buzzword or tool to be defined in terms of the dimensions illustrated in Figure 1.

Figure 1: Summary of BI Taxonomy

Let's test the taxonomy using business activity monitoring (BAM) as an example. A BAM product is used for making operational decisions and actions. It consumes detailed business transaction data and creates summarized business metrics and alerts. The latency of both consumed and created data is real time or near-real time. As data flows through a BAM application, it is cached in an in-memory store and may be made persistent in a low-latency data store.

A BAM application uses event monitoring to track business transaction data, an analysis engine to calculate business metrics, a rules engine to produce alerts and a dashboard builder to display metrics. Business rules used by BAM applications include transformation rules to integrate and filter transaction data, analysis rules to create metrics, context rules to put metrics into a business context and exception rules to raise alerts. A BAM application may also employ a data quality management tool to validate in-flight transaction data and a portal to provide collaboration and display dashboards and alerts.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access