Some corporations are multi national. This means that there is a headquarters and then there are outlying organizations around the world. These multi national corporations face a unique problem. These corporations need two types of data local data and global data. Local data is data typically found in a data warehouse. But global data is different. Global data is data that has been taken from the local environment and has been recast into a global mode. Fig global.1 shows a global map. A corporation has its headquarters in Singapore. The corporation has outlying offices and plants in Denver, Colorado, USA, in, Brazil, and in Prague, Czechoslovakia. Fig global.1 Consider the challenge of managing global data across many branches of the corporation It is a temptation to say that the problems of creating global data is as simple as installing SAP in each location, as seen in Fig global.2.
Fig global.2 While making every country have SAP software solves some problems, there are other problems that are not solved Indeed, installing SAP in each location solves many problems. But it doesn t solve the problem of global data. The reason is that each implementation of SAP worldwide is different at the semantic level from each other implementation. Fig global.3 shows that different implementations of SAP follow different business practices. USA Singapore Fig global.3 The problem is that the SAP USA implementation is different than the SAP Singapore implementation Fig global.4 shows that what is needed to resolve the different implementations of SAP worldwide is rationalization of data as data is passed into a global data warehouse.
A global data warehouse Fig global.4 A global data warehouse and several local data warehouses Local data warehouses In order to accomplish such a rationalization, first a global data model is created at headquarters. This model is shown in Fig global.5.
Fig global.5 First the global data model is created Fig global.5 shows that a global data model is created. The global data model reflects only the data that is needed globally. Usually this model is remarkable in that it is small and simple. A model with typically 30 or 40 attributes is normal. In many cases the model relates to financial data. In other cases the global model encompasses customer information. In any case the global model is far more simple than the standard data model. One feature of the global data model is a definition of each of the attributes as the corporation wishes those attributes to be defined. Fig global.6 shows the shipment of the global data model to the local environments.
Fig 12.6 next the global data model is sent to each of the local sites After the global model is created, then the model is shipped to each local environment. At each local environment the local data is mapped to the global data. In many cases this means a recalculation or a reinterpretation of local data. In any case, a clear mapping of how local data will be used to satisfy the need for global data is spelled out. Fig global.7 shows the local mapping of data into global data.
Fig global.7 Then the global data model is mapped against the local data. Each local mapping is different from each other local mapping After the local mapping is created, the next step is to translate the local mappings into physical tables. Fig global.8 shows such a translation process.
Fig global.8 Now the mapping is turned into programs that access the local data warehouse and then populate data ready to go to the global data warehouse Of course ETL can be used in this local creation of tables and data bases. After the local tables have been created, they can then be shipped to the global data warehouse. Fig global.9 shows such a shipping.
Fig uns.9 Finally the local data that has been mapped is sent to the global data warehouse The process of creating local renditions of global data and the process of shipping the local data to the global data warehouse is then institutionalized and scheduled. The shipment may occur daily, weekly, monthly or at any other schedule that makes sense. Once created and populated there is now a global data warehouse and a series of local data warehouses. And in each warehouse there is a different perspective of data, as seen in Fig global.10.
Fig global.10 Different people with different perspectives use the global data warehouse and the local data warehouses The process of receiving data is an around the clock exercise. Fig global.11 shows that data may be received at the global data warehouse around the clock.
Fig meth.11 The process of receiving data is an around the clock exercise