Building the Data Warehouse

Скачать в pdf «Building the Data Warehouse»


A related yet separate issue not addressed in this chapter is global operational data. Thus far, this chapter has assumed that every local site has its own unique operational data and processing. However, it is entirely possible that there is a degree of commonality between the operational systems found at each local site. In this case, some degree of corporate operational data and processing might be desirable. For example, some customers may need to be handled globally, such as large multinational corporations like Coca-Cola, McDonalds, IBM, and AT&T. Pricing considerations, size-of-order considerations, and delivery considerations may be treated differently globally than similar decisions are treated locally. In the case of global operational processing, that global operational data merely becomes another source of data into the global data warehouse. But there is still the distinction between operational data and DSS informational data.


Underlying the whole issue of the distributed data warehouse is complexity. In a simple central data warehouse environment, roles and responsibilities are fairly straightforward. In a distributed data warehouse environment, however, the issues of scope, coordination, meta data, responsibilities, transfer of data, local mapping, and more make the environment complex indeed.


One of the major considerations for a global data warehouse is whether the data warehouse should be built centrally or globally. While tempting to say that a global data warehouse should be designed and built centrally, doing so is patently a mistake. With a centralized construction of a global data warehouse, there is—at best—only a marginal local buy-in to the global data warehouse. This means that the definition of the mapping between the local systems and the needs for global data is done centrally, not locally. To succeed, there must be local management and control of the mapping process. Stated differently, the single most difficult part of the building and population of the global data warehouse is the mapping of the local data to the global data. And this mapping cannot be done centrally; it must be done locally.

Скачать в pdf «Building the Data Warehouse»