Temporal Databases and Data Warehousing

Temporal databases and data warehousing are two separate areas which are strongly related: data warehouses are the commercial products that require temporal database technology. Naturally, most other database products are amenable to temporal database technology, too. Regarding the market perspectives, however, one has to assume that it will be mainly data warehouses that adopt the techniques that have been and that will be developed by temporal database researchers. In this section, we want to elaborate the connection between data warehousing and temporal databases in some more detail.

A data warehouse (DW)  integrates information from many, possibly heterogeneous, databases into a physically separated database and makes this information available to analysis [Inmon, 1996]. Figure 2.5 illustrates this concept. The purpose of the analysis might be, for example, to provide the management of a company with information on trends and facts that are required for taking strategic decisions.

  

Figure: The concept of a data warehouse. 

Trend analysis can go along many dimensions, the most important of which is time. It is used to detect certain characteristics in the evolution of data, e.g. over time or over various geographic regions or over product lines. In the case of temporal evolution, this means that a data warehouse is very often required not only to hold a reformatted subset of current operational data, e.g. sales figures, but also a history of this data. This is nothing other than a historical database , a special case of a temporal database [Sarda, 1993]. For that reason, Inmon says that a ``salient characteristic of the data warehouse is that it is time variant''. Furthermore he comments:

• Data warehouses are required to hold data of the last 5 to 10 years whereas operational databases require a 60 to 90 days time horizon.
• ``Operational databases contain current value data - data whose accuracy is valid as of the moment of access. As such, current value data can be updated. Data Warehouse data is nothing more than a sophisticated series of snapshots, taken at one moment in time.'' This comment corresponds widely with the concepts of the time cube and physical vs. logical deletion that were introduced in section 2.3.
• ``The key structure of operational data may or may not contain some element of time [...] The key structure of the data warehouse always contains some element of time.''

• Many data warehouse vendors claim that their products are capable of processing historical data / information. Examples of such vendors are RedBrick [Red Brick Systems, 1995a], Informix [Informix Inc., 1995], Prism Solutions [Prism Solutions Inc., 1996], Oracle [Oracle Corp., 1996].
• Researchers from SAP claim that data warehouses must have the ability to meaningfully link and cross-reference data ``applying time-related criteria''. Furthermore they state that one salient feature of DW data management is the ``time-variant data organization'' [Heinrich and Hofmann, 1996].
• Red Brick Systems call their product RedBrick Warehouse VPT, in which `T' stands for the fact that it provides ``time-based data management'' [Red Brick Systems, 1995c].