Читать книгу Applied Microsoft Business Intelligence - Sarka Dejan - Страница 13

With the users and goals identified, the time has now come to perform one of the most difficult steps in the project – identifying the data sources. Believe it or not, a person or group of people from the expected end users are the best source for this task. IT data sources often reference data that resides only in systems that IT manages with no regard to data hosted by a department, branch office, or third party. For a business intelligence project, this is not typical. You must perform an exhaustive search-and-discovery process with as much involvement from any stakeholders that are willing to assist. You may host these sources within or outside an organization.

Regardless of whether the data is internal or external, you should carefully perform discovery to ensure that you have included all relevant data in the project. This process often requires several iterations. During report development, someone may recognize that data is missing due to an oversight. As a result, you'll need to modify the ETL process to include the new source.

Using Internal Data Sources

While you source the majority of the data from traditional IT-managed relational databases, you'll always have some data managed and maintained outside the IT department. This data is likely stored in spreadsheets, Access databases, comma-delimited files, text files, or other file types, and they may reside on someone's desktop or laptop. Often vital, these datasets contain small nuggets of information that can cripple the project if they are not included.

End users may also manage other internal data sources, such as SharePoint lists or third-party applications that came with database backed during installation. In the case of the latter, the hosting department does not even realize what they have installed. In some cases these are common back ends, and others may require custom drivers to access the data.

Using External Data Sources

The data may also come from an external source via a web service, an OData feed, or even a hosted RDBMS (SQL Server or Oracle). If the source is a web service or OData feed, the data is typically accessed via a web URL. The consumption, on the other hand, may require some custom interface that parses and displays the data in a fashion meaningful to end users. Developers may overlook this data because no one on the team knows it exists; the same may apply to the hosted databases. As a result, this further heightens the need to involve end users because they may be the only people who know it's there and needed.