Product Administration Guide > Select a Configuration Method >

Consider the Nature of Your Data

To select between the two methods, consider whether your data is dense or sparse.

Dense Data

When your data set is "dense" and relatively small, consider deploying a browser-based application. This model makes it easy to apply business rules that manage the different combinations of data.

Table 6 is an example of dense data. In this example, a set of shirts is available in a variety of colors and, except for yellow, in all sizes. Yellow is available only in small. The combinations of data are easily entered in Configuration tables in Interactive Designer.

Table 6. Sample Browser-Based Data Set
Color
Size
Red
*
Green
*
Blue
*
Yellow
Small

At runtime, users can freely explore options for a given configuration, make selections, and determine what features are most important to them, not necessarily what is valid based on the first or top level selection. As exception messages are presented for invalid selections, it is relatively easy for users to get back to a valid state, because of the relatively small number of exceptions and relatively "dense" data.

Alternatively, within this deployment model, users navigating a large and sparse data set may spend more time than desired trying to identify a valid state. To avoid this situation, modelers can build in constraints to minimize the number of "invalid states." However, you will need to weigh the trade-offs between the time spent building in constraints versus modeling in a constraint-based environment, particularly as product complexity increases. In this situation, consider using the Server-Based model explained in the next section.

Sparse Data

When your data set is large and "sparse," consider deploying a server-based application.

Table 7 shows a simple model where the data is more sparse and contains multiple, complex exceptions. This type of model is more efficiently expressed as a set of constraints.

Table 7. Sample Server-Based Data Set
 
Feature X
Feature Y
Feature Z
Product A
 
Yes (*1)
*
Product B
Yes (*2)
 
*
Product C
 
 
*

*1 - only if x>y and not combined with A
*2 - only if A and B are combined


 Product Administration Guide, Version 7.5 
 Published: 18 April 2003