Benefits of Data Warehouse (Part2)
Roles of data warehousing – benefits and risks
To better illustrate different scenarios of using data warehouse, simple model of data warehouse is used of an insurance company. Defined are four dimensions: insured person, time, services, point of sale. Measure is fee billed. In this example, a simple datamart (DM) replaces a very large number of static reports (Figure 1). Datamart will simplify business for persons engaged in regular (eg monthly) reporting at any level.
Figure 1. An example of a simple Datamart.
DM replaces also reports that are typically performed only once and never again use the (for example who are insured in place A and have bought XXX service at all retail outlets in place B, or for example which service makes the most money per quarter) Often for various reasons, which are needed within the company, reports that appear to be generated very rarely or only once. These are in fact all statements that do not belong to a set of reports that are standard and serve as leading indicators of business structures in the company, but serve for example to market research or development services, for example at some point company decides to cancel certain insurers, which often damage the cars to raise monthly insurance.From this point of view data warehouse model can be determined with the help of tools for making statements closer to the person who must report. User very quickly and efficiently create reports, which frees up considerable human resources in the IT department. Data Warehouse has several parallel roles (in text below) but mostly during construction of DWH only one role is being favored. Decision about favorite role is based upon recognized need for improvement of the existing information systems companies.
More information in Data Warehousing PDF.
1. Controlling role of DWH
During the process of extracting, data can be controlled and filtered (data cleansing). Data that failed to pass filters are candidates to be the data - garbage, errors. Take for example the simplest filter that checks is entering date of payment of any party in relation with the current year. In most production systems during the year 2000 for various reasons year 1900 was entered.
Filters can be much more complex, some of them may be subject to the business process (non-existent customer can not pay) and by statistics. For example, it is likely an error if one day we have over 120,000 USD. at a particular point of sale where Y is normally charged 10.000 USD dailly. Record of 110.500 USD for application is acceptable, but employees can enter errors and this the case of manual error. Therefore it makes sense to build a warehouse alert system that tells us that sale of the location Y is outside of the statistical framework.
Figure 2 presents schematic process of extraction of data through the filter. These filters are installed in the source code extraction.
Figure 2 The process of extraction of data through the filter.
2. Strategic Role
Consider for example that the data warehouse insurance data has back up of fifteen years. The amount of data can contribute to defining its knowledge of enterprise business processes with datamining tools. It is possible to discover simple things such as which services over the past n years were the most profitable in the long term and which generated losses. Some services are analyzed through a number of years to see whether the investment in the service was worthwhile. On the other hand, sometimes users can find out more complex data that can not be assumed before analysis, like for example which streets are highly at risk of burglary and which are safe. With this information insurance fee will be settled for clients. On the other side this example shows how DWH role falls again under the control role, because it can identify workers who often make mistakes when entering data (if the DWH contains dimension for each employee entered data).
3. Hardware accelerator
Reporting and data input on the same particular server at a same time slow down operative work of server. Therefore, it is necessary to purchase a new server that is used for reporting only, to be specific for reporting processes only. Fact is that if the server on which the data are entered only will not have any more complaints from users who import data if reporting function is removed from server. Usual complaints will not be any more on slow work that sometimes lead to serious delays in the work. Tasks of reporting are removed on the server which is usually extension of data warehouse.
Data Warehouse as software structure even if it is placed on only one server speeds up the process of reporting due to functional customization for needs of reporting. Because of this feature data warehousing is functioning as hardware accelerator simply due to fact that existing structure of production systems are released of demanding reporting process. Before the introduction of data warehousing statistical reporting usually slowed performance of production systems.
Benefits of Data Warehouse (Part1)