Datawarehouse
Datawarehouse is the component exchange of the decisional system. It has vocation to store in a centralized and transverse way, the whole of the data necessary to piloting. Datawarehouse contains all information considered to be relevant for the decision-making aid, extracted beforehand, then harmonized.
According to Bill Inmon, in 1994: “Datawarehouse is a collection of data directed prone, integrated, not birds and historized, organized for the support of a process of decision-making aid”.
Directed prone
The objective of Datawarehouse is decision making around the major activities of the company. In Datawarehouse, the data are thus structured by topics in opposition to those organized, in the systems of production, by functional process. The interest of this organization is to have the whole of the useful informations on generally transverse subject with the functional and organisational structures of the company. One can thus pass from a vertical vision of the company to a transverse vision much richer in information.
Integrated
In an ideal world, the information systems sources (systems of production) are homogeneous and the company has the perfect knowledge of all codings of which it has need to benefit from its informational capital. In reality, the data, resulting from various applications of production, exist in different forms. It is then a question of integrating them in order to homogenize them and of giving them an one way, comprehensible by all the users. The required transversality will be all the more efficient as the information system will be really integrated. This integration requires a strong standardization, a good management of the reference frames and coherence, a perfect control of semantics and rules of management applying to the handled data. It relates to internal data but also external data which pose problems because their coding and their level of detail different from those of the internal data. It is only at the price of a “successful” integration that one can offer a homogeneous and coherent vision company via his indicators. This supposes that the information system of the company is structured, controlled well already well, and profits from a sufficient level of integration. If such were not the case, the quality of the data can prevent the good implementation of Datawarehouse.
Not birds (nothing is lost)
In order to preserve the traceability of information and the decisions taken, the information stored within Datawarehouse should not disappear. The same request launched several times, and this in months of interval, on the same population must restore the same results. Thus, since a data was qualified to be introduced within Datawarehouse, it can neither be faded, neither modified, nor removed (or in any case in on this side certain time of purging). It becomes, in fact, recipient of the history of the company. This characteristic differs from the logic of the systems of production which very often update the data by cancel and replace with each new transaction.
Historisées
Not-volatility allows the historisation. From a functional point of view, this property makes it possible to follow in time the evolution of the various values of the indicators to analyze. In fact, in Datawarehouse a reference frame of time is necessary. It is the axis time or period.
Organized
The data are organized in a specific way for the decision-making aid.It is useful to specify that Datawarehouse is conceived to contain the data in adequacy with the present needs and future for the organization, and to answer in a way centralized all the users. Thus, there is no rule purist which applies as regards storage, nor of single modeling: the datawarehouse can contain certain detailed informations, resulting from the sources of production, necessary to a need for recurring operational piloting, just like of the tables of facts, ready with employment.
Directed prone
One says that Datawarehouse is directed “Prone”, in answer on the various subjects of the company which it is supposed to prepare with the analysis.| Random links: | Anne of Armagnac | (503) Evelyn | Jon Christensen | Breather | Philippe Charlier |