Performance optimization for IBM Cognos Analysis Studio
If the length of time required to perform interactive analysis is unacceptable, or if extensive navigation is required to find information, evaluate the way that you built the analysis. Avoid including information in the analysis that is not required to answer your business question. For example, if you are interested in specific members in a hierarchy, add only the required members instead of the root member. This reduces the amount of navigation required to find the relevant information, or the need to use other methods to limit the data view.
Analysis Studio is most effective when you perform analysis on data that displays over a relatively small number of pages. Smaller data sets are ideal because Analysis Studio is optimized for real-time interactions with the data as opposed to simply creating output. The structure of some large data sets might result in displays that span such a large number of pages that performance is unacceptable or navigation is difficult. For example, a data set might include thousands of rows at one level with no values for the majority of the rows. With this data set, the analysis is difficult to navigate due to the large number of pages required to display results. Also, applying suppression might take an unacceptable length of time. Other IBM Cognos studios, such as IBM Cognos Report Studio Professional Authoring, are more appropriate tools for working with these types of large data sets.
The first step to avoid usability and performance issues with reports is to have a clearly defined business question. This will help you to identify the specific data required to build an analysis. Next, understand the options available for determining the data to include in the report. Test a variety of layout and data control options for improving usability and performance.
Suppression
If the length of time required to apply zero or null suppression is unacceptable, test an alternate method of limiting the view of the data.
- Use selection-based suppression.
- Reduce the number of members that appear on the crosstab edge. This is especially effective when the majority of cells have null values.
- Show a limited number of members based on data value.
- Use a filter to show only members that meet a measure value threshold.
Nesting
To avoid requests for result sets that are too large for real-time interactive analysis, you cannot nest more than three levels from the same hierarchy in IBM Cognos Analysis Studio. If there is a requirement to view more than three nested levels from the same hierarchy, it is best to create a report in IBM Cognos Report Studio Professional Authoring. You can set up drill through to analyze the area of interest in IBM Cognos Analysis Studio.
Levels
The method that you use to add levels to an analysis determines the way in which summary values are generated.
When you insert a level by selecting the parent, summary values are obtained directly from the data source. OLAP data sources typically optimize rollups to improve performance.
For some insertion methods, summary values are calculated when you add the level. For example, with the following actions, summary values are calculated. The summary value does not come directly from the data source.
- Right-click a member from the Insertable Objects pane, and click Insert Level.
- Select a member from the Insertable Objects pane. In the Information pane, select the level and drag it to an edge.
This behavior is used to ensure that the summary is correct even when the displayed members do not roll up into an accessible parent member. The dynamic calculation of summary values can have a negative affect on performance.
In some cases, summary values are always calculated regardless of the method used to add the level. For example, summary values are calculated when you use a top or bottom filter.
Workflow guidelines
The following workflow guidelines will help you avoid usability and performance issues.
- Add members by dragging the root member to a crosstab edge. If you are interested in only some members in a level, drag individual members to the crosstab edge.
- To limit the view to a slice of the cube, such as a specific year, drag the item to the context filter area. Do not use members from a hierarchy that appears on a crosstab edge as a context filter.
- To remove sparse data in the analysis, apply suppression. If it takes an unacceptable length of time to apply general suppression, review the layout of the analysis to identify options for reducing information before retesting suppression or other options for limiting the view of the data.
- Change the layout to test alternate options for showing a lower level of detail: drill down, expand, and down a level.