Optimizing the Performance of IBM Cognos Analysis Studio

Resolving The Problem

By design, IBM Cognos Analysis Studio is optimized for slicing and dicing a meaningful view of your data at a time. As such, the best approach to analysis is to reduce as much as possible before each expansion step. In order to do this, you should try to ensure the output is small enough that you can see it all without too much navigation. Other studios are better suited for high volumes of detail rows that span many pages.

The following is a common usage pattern that is not recommended::
1) Drag something from the Insertable Objects pane to the rows or columns.
2) Drag specific members from the same hierarchy to the Context filter.
3) Since the result is sparse, turn on suppression to make it dense again.

The following instructions are a general recipe for success:
1) Start an analysis by dragging the root member from the hierarchies of interest (which will automatically show the first few members, along with an indication of whether there are more).
2) If you are only interested in specific members, either drag them on or select them in the analysis and Keep them.
3) If you are only interested in a slice of the cube that involves hierarchies that don’t need to be displayed in the crosstab/chart, add them to the Context filter area.
4) If the result is extremely sparse at this point, apply suppression.
5) If the suppression step does not return a usable result in a few seconds, cancel or undo it, and look at the layout and see if you can think of any way to reduce the size of the un-suppressed result in a way that gets rid of the More indicators. The section below provides options for improving suppression performance.
6) Now if you need to see a lower level of detail, options to do so (in order of decreasing performance) are:
• drill-down
• expand
• down a level
7) If this makes the result too big to look at easily or takes too long, try to refine it further using the techniques above. Continue this refine/navigate/expand cycle until you have the answer you need.

At some point, you may reach the limit of the amount of data that you can reasonably look at, or that the system can reasonably handle. At that point, you should consider either re-thinking the business question you are trying to ask (based on the results you have seen so far), or using a reporting tool instead of an analysis tool.

Suppression
If it’s taking a frustrating amount of time to suppress the zero and/or null values of your analysis, consider the following options:
1) Use selection-based suppression instead of general suppression. Steps are at the bottom of the following page
https://www.ibm.com/support/knowledgecenter/SSEP7J_10.2.2/com.ibm.swg.ba.cognos.ug_cr_pps.10.2.2.doc/c_id_pps_spprss_ec.html

2) If the user knows which members will have data they care about, explicitly place or keep only those on the crosstab edges instead of all members, particularly when the vast majority of cells will be null.

3) Show only the top N members. More on this here https://www.ibm.com/support/knowledgecenter/SSEP7J_10.2.2/com.ibm.swg.ba.cognos.ug_cr_pps.10.2.2.doc/t_id_pps_top_bottom.html#id_pps_top_bottom

4) Filter members to show only those whose measure value is above a certain threshold. More on this here
https://www.ibm.com/support/knowledgecenter/SSEP7J_10.2.2/com.ibm.swg.ba.cognos.ug_cr_pps.10.2.2.doc/c_id_ug_cr_pps_limitng.html


Nesting
In order to avoid requests for result sets that are too large for real-time interactive analysis, IBM Cognos Analysis Studio prevents users from nesting 4 or more levels from the same hierarchy. If there is a requirement to view 4 or more nested levels from the same hierarchy, the recommendation is to do so in a IBM Cognos Report Studio Professional Authoring report and then, possibly via drill through, analyze the area of interest in IBM Cognos Analysis Studio.

Levels
In most dimensions, every non-leaf member is the roll-up (i.e., the natural summary) of its children. As opposed to having summary values computed in real-time, if you use the parent member this value will come directly from the data source. OLAP data sources typically optimize these rollups for good performance.

In IBM Cognos Analysis Studio, some gestures including the following will add a level to the analysis:

a) Select a member from the Insertable Objects pane, right-click, and select Insert Level.
b) Select a member from the Insertable Objects pane and in the Properties pane below it select the Level and drag it onto an edge.

In any of these cases, if totals are enabled they will be calculated at that time rather than returning the value of the parent member. This is done in order to ensure that the total is correct even when the members that are displayed do not roll up into an accessible parent member. Avoiding these gestures, in favor of using the desired parent member, will avoid the associated performance costs. You can’t use this technique if you need a summary of set of members that is not a complete set of children (e.g., a TopCount set). In this case, the summaries must be computed on demand.