Often, multiple applications will end up using the same data. Why retrieve it multiple times? Retrieve it once, build a QVD, and then use the QVD in the user applications. And what if the database you're reading only has inserts, so old data never changes? In that case, it would be much more efficient to buffer the old data in a QVD, and only read the new inserts. A QVD also allows the user applications to not care about the details of your data source. You could change from reading an Excel file to reading data from a database, and as long as the user applications were reading QVDs, you'd only have to modify a single application.
There are surely cases where it makes sense to just query tables directly. But there are also plenty of cases where it makes sense to put in a QVD layer.
Often, multiple applications will end up using the same data. Why retrieve it multiple times? Retrieve it once, build a QVD, and then use the QVD in the user applications. And what if the database you're reading only has inserts, so old data never changes? In that case, it would be much more efficient to buffer the old data in a QVD, and only read the new inserts. A QVD also allows the user applications to not care about the details of your data source. You could change from reading an Excel file to reading data from a database, and as long as the user applications were reading QVDs, you'd only have to modify a single application.
There are surely cases where it makes sense to just query tables directly. But there are also plenty of cases where it makes sense to put in a QVD layer.
