Information Modeling Prerequisites

The information model is the core deliverable of the technical blueprint development phase of an SAP BW implementation project, directly following the business strategy development/validation and the requirements-analysis steps in the overall project plan. The deliverables of those preceding steps providing the prerequisites for developing the information model include the following:

    *  Business intelligence strategy — Often neglected is a business intelligence strategy directly aligned with the overall business strategy of an organization; it defines what business information is needed (requirements and priorities) and how to technically deliver it (overall architecture and implementation approach). From a business perspective, more advanced business intelligence strategies not only incorporate the defining of core metrics (historical reporting) but also the process of explaining them (descriptive modeling), sensing them (predictive modeling), and acting on them (prescriptive modeling). From a technology perspective, more advanced business intelligence strategies incorporate the latest innovations and related architectural trends such as Web services and SOA.

    *  Business processes and analysis requirements — This refers to the complete set of detailed analysis requirements for the business processes in focus, including the key performance indicators (KPIs) used to monitor these business processes. KPIs may be anything from simple key figures describing a business event (such as revenue in a sales transaction) to complex derived key figures, eventually based on restricted key figures and calculated key figures (such as a customer lifetime value).

    *  Relevant entities — As a result of the requirements- analysis phase, strong and weak entities relevant to business processes are identified. Strong entities are those such as customer, product, account, and time, which are directly related to a business event; weak entities are related to a strong entity and describe the business event in more detail. Examples of weak entities include ZIP code, product groups, and other classifying information. In SAP BW, both strong and weak entities can be used as characteristics in an InfoCube and have master data attributes, texts, and hierarchies assigned to them.

    *  Attributes of relevant entities — Attributes provide additional information about an entity (such as address information or the customer group for a customer entity). Attributes available for display can be made available for navigation in queries (navigational attributes). The concept of navigational attributes is important for the information modeling process, because it allows handling the slowly changing dimension phenomenon in an elegant way (more on this later in this chapter).

    *  Hierarchies of relevant entities — Hierarchical relationships between different objects of the same kind (for example, cost center hierarchies) or between different objects of different types (for example, regional hierarchy grouping customers) are important for analysis purposes. Hierarchical relationships can be modeled as hierarchies or as (a collection of) attributes. Choosing hierarchies allows for more flexibility but is technically more expensive.

    *  Dependencies between relevant entities — Dependencies between relevant entities have a major impact on the information model and on the transformations required to populate an information model with data. This is not restricted to technical information captured in an entity relationship model but reaches out to how changes in business processes (such as reorganizations) actually affect the relationship of entities (for example, the relationship between a product group and a product or an employee and a cost center).

    *  External information — Measuring the performance of a company means comparing it to something else (for instance, to internal information such as budgets, plans, and previous year results but also to external information such as market shares or demographic data). External data sources (for example, address databases and credibility databases) can also be used to verify and enrich information in the data warehouse. While the actual origin of information is not relevant for the information modeling process, information from outside the enterprise can be very valuable in terms of benchmarking the development of the enterprise against industry trends.

    *  Data Volume — Information about the data volume must be considered for all entities from two different perspectives. From a static perspective, you need to know how much data will have to be stored and maintained; from a dynamic perspective, you need to know how many data records you need to process in a certain period of time.

    *  Granularity — The granularity of information defines the level of detail of the information provided. It may range from a line-item or schedule-line level in sales transactions, to highly aggregated management information. There is, of course, a trade-off between performance and granularity: The higher the granularity of a reporting object, the larger the data volume and, obviously, the longer query execution will take. Granularity typically differs among the different levels of the data warehouse architecture.

    *  Timeliness — The timeliness of transaction and master data is determined by the frequency in which InfoCubes, DataStore Objects, and master data are updated. In many cases, information in a data warehouse is updated on a daily, weekly, or monthly basis. Some applications, however, require more frequent updates or even real-time information (for example, Web-log analysis).

    *  Functional requirements — Functional requirements (as opposed to business requirements), include a description of user interfaces, systems to be available for extraction, Web servers for Web reporting, hardware, and software.

    *  Nonfunctional requirements — Nonfunctional requirements include (but are not restricted to) system properties, such as environmental and implementation constraints; performance; platform dependencies; maintainability; extensibility; and availability.

    *  Information delivery requirements — Information delivery is not restricted to making reports available to end users. In many applications, information is also exported to external companies (for example, market research transferred to ERP or legacy systems and planning results imported back into an operational system to control production).

    *  Usability — Although information modeling has little to do with designing front-ends and front-end tools, it still has an impact on the usability of the information system. A transparent, easy-to-understand information model helps end users understand predefined queries and define their own ad-hoc queries.