C2090-632 IBM Cognos 10 BI Metadata Model Developer Practice Questions

The core benefit of a layered approach is abstraction. The Presentation Layer, which is what report authors use, is built upon the Business Layer. The Business Layer, in turn, is built upon the Physical Layer. If a table in the database is renamed (a physical change), the modeler only needs to update the Physical Layer and remap it to the Business Layer. The Presentation Layer and all the reports built on it remain unchanged, dramatically improving maintainability and reducing the impact of source system changes.

The #...# syntax indicates a macro expression. The $ sign within the macro is used to reference a session parameter by its name. This expression will be evaluated at runtime, substituting the value of the 'userCountry' session parameter for the currently logged-in user into the WHERE clause of the query.

'Limited Local' processing is a hybrid approach. Cognos will push as much of the query execution as possible to the underlying database (e.g., joins, filtering, sorting, standard aggregations) to leverage its power. However, if the query uses a specific function that the database does not support natively, Cognos will process that part of the query locally on the Cognos server after retrieving the base data.

Renaming items to use clear, unambiguous business terminology is one of the most important steps in creating a user-friendly business view.

Folders help organize a potentially large number of items into logical, collapsible groups, making it much easier for users to find what they need.

Screen tips act as tooltips in the reporting studios. They are an excellent way to provide in-context help and documentation directly to the end-user without cluttering the model structure.

A chasm trap occurs when two fact tables are joined through a common dimension. Cognos resolves this by generating a 'stitched query'. It queries each fact table separately against the common dimension(s) and then 'stitches' the results together using a full outer join on the dimension attributes. The model must be configured to allow this, typically by ensuring facts are only joined to their corresponding dimensions, which forces Cognos to recognize the pattern and generate the correct multi-fact query.

Creating a model query subject to consolidate the snowflake schema into a single, denormalized dimension is the best practice. This simplifies the model for report authors, improves query performance by pre-defining the joins, and makes the business view more intuitive.

A fan trap occurs when a one-to-many join links to another one-to-many join from the central table. The correct way to resolve this is to create query subjects that isolate the join paths. Cognos can generate a stitched query to correctly bring the data together. Often, creating a shortcut to the relationship or creating separate query subjects for different contexts is the standard solution.

Isolating metadata from each data source into its own namespace is a fundamental best practice. It prevents naming conflicts, clarifies data lineage, and simplifies maintenance.

A layered modeling approach involves creating a separate namespace for the business view. This abstraction layer uses shortcuts to the physical namespaces, providing a stable and user-friendly interface for report authors that is insulated from changes in the underlying sources.

This is the correct approach. By applying the filter to the security table DoctorFacilityAccess, you restrict its contents to only the facilities associated with the logged-in doctor. Because this table is joined to the fact table (likely through a facility dimension), the filter propagates through the model, correctly securing the PatientEncounters data. Applying the filter to the security mapping table itself is the most efficient and maintainable design.

This is true. Framework Manager provides this behavior as a convenience for modelers. When it detects that a numeric data item is being reclassified as a Fact, it assumes the most common aggregation function, Sum, should be applied. The modeler can then change it to another function like Average or Count if needed.