Datamart

A datamart is a store data. As the Datawarehouse it is a Entrepôt of data, but specialized for a trade or a subject donné.
In general, it is located downstream of a more total Datawarehouse and is fed starting from this Datawarehouse, of which it constitutes to some extent an extract.
The datamart is a specific point of view, determined by functional criteria trade. The datamart is conceived according to a identified Besoin for business and for a clearly definite range. Thus several datamarts could be conceived to answer each one a Besoin différent.
The datamart is a decisional database structured and formatted according to a precise trade or of a particular Usage (datamart commercial, datamart financial,…). Information is prepared there to be “consumed” such as it is: it is a structure more specific than a Datawarehouse which remains, him, with more general vocation and more transversal taking into consideration trade various of the company.

The Entrepôt of data of the datamart type has a Structure of data:

multidimensional is stored in a cubic or hypercube: M-OLAP, R-OLAP… (in particular when it is about Donnée S aggregate)
is represented by a Data model in star or in flake in a relational DBMS (in particular when it is about Donnée S elementary or unit nonaggregate).

The datamart is used with the Reporting and the decisional analysis, in end of the chain of the decisional Informatique.

In one warehouse dated correctly conceived, dated them marts is fed and brought up to date by a database centralized on the scale of the company, container of the precise and up to date data. This is ideal, but very often dated them marts in are developed outside and independently of any architecture of warehouse dated. The profits of implementation realized in the short run at the time of the constitution of dated marts " indépendants" result quickly in difficulties which perdurent in the long run. Marts dated do not present any inherent main issue. In terms of performances, they are of considerable interest for repetitive and well-known interrogations. The problems actually occur only when they constitute the single means of access to the data for the end user. In this type of approach, all the environment of support of decision worked and is adapted starting from a whole of known questions and answers foreseeable, without taking account of the evolution of the interrogations of the company. In terms of management, aspect which constitutes one of the main arguments of sale of dated marts, the answer is not as simple as one could think. The argument is indeed the following: one dated mart is a whole reduced and concise of information joined together in a single place and thus easier to feed and bring up to date. This argument which is justified one in the case of dated mart quickly single émiette as soon as marts dated them multiply. The management of several dated marts independent not supplied with one dated warehouse centralized matches costs. The loads of implementation of complex tasks of synchronization and development of functions of interrogation not limited to one dated mart specific quickly carry it on the savings made at the time of implementation the initial.

Datamart

See also

Other articles