Summary
In this chapter, we discussed the BW star schema and analyzed three InfoCube design alternatives for the same business scenario described in Chapter 1. We also briefly discussed two other InfoCube design techniques: compound attributes and line item dimensions.
Key Terms
| Term | Description |
|---|---|
| SID | Surrogate-ID (SID) translates a potentially long key for an InfoObject into a short four-bytes integer, which saves I/O and memory during OLAP. |
| Display attribute | A display attribute provides supplemental information to a characteristic. In our example, IO_CUSTNM (Customer name) and IO_CUSTAD (Customer address) are display attributes of IO_CUST (Customer ID). |
| Navigational attribute | A navigational attribute indicates a characteristic-to-characteristic relationship between two characteristics. It provides additional information about a characteristic and supports navigation from characteristic to characteristic during a query. In our example, IO_SOFF (Sales office) and IO_SREG (Sales region) are navigational attributes to IO_SREPN1 (Sales representative ID). |
| Compound attribute | A compound attribute differentiates a characteristic so as to make it uniquely identifiable. For example, if the same characteristic data from different source systems means different things, then we can add the compound attribute 0SOURSYSTEM (Source system ID) to the characteristic. 0SOURSYSTEM is provided with the Business Content. |
| Time-dependent hierarchy structure | The time-dependent hierarchy structures consist of nodes or leaves that are time-dependent. The hierarchy itself is not time-dependent. In our Chapter 3 example, the Denver office is listed twice in the hierarchy: Before January 1, 2000, it belongs to the Midwest region; on and after January 1, 2000, it belongs to the West region. |
| Time-dependent entire hierarchy | A time-dependent entire hierarchy is a time-dependent hierarchy whose nodes and leaves are not time-dependent. The Chapter 7 example includes two hierarchies. One hierarchy is for the period January 1, 1000, to December 31, 1999; during this period, the Denver office belongs to the Midwest region. The other hierarchy is for the period January 1, 1999, to December 31, 9999; during this period, the Denver office belongs to the West region. |
| Line item dimension | A line item dimension in a fact table does not have the dimension table shown in the simple star schema. Rather, it connects directly with the SID table of its sole characteristic. |
| Granularity | Granularity describes the level of detail in a data warehouse. It is determined by business requirements and technology capabilities. |
Next…
We will discuss aggregates and multi-cubes. Aggregates help improve query performance. Multi-cubes support cross-subject analysis.