PA to PAaaS upgrade considerations and differences

Cognos Analytics leverages the Planning Analytics Engine in Planning Analytics as a Service. As a result, MDX queries are handled entirely by Planning Analytics and queries conform to their standard. The local MDX engine and MDX caches in Cognos Analytics are no longer used.

Considerations

If you plan to upgrade from a Planning Analytics connection, consider the following differences between the two connection types:

  • The query service requires less memory and sends fewer queries to Planning Analytics when it runs queries. Consider reducing the memory for the query service.
  • Settings related to running MDX in Cognos Analytics are no longer used. This includes priming query thresholds, the IsUnderFed cube property, and the allowSuppressingStringMeasures and allowFilterOptimizationForUnderFedCubes properties. These are either no longer required or they have defaults that are defined in Planning Analytics.
  • You don't need to change packages or data modules. If you use Cognos Administration to create the connection, the existing IBM Planning Analytics data source connection type can be used with the port left empty and SSL check box unchecked.

Differences in output

After you upgrade from Planning Analytics, you may notice some of these differences in your output:

  • Formatting that is defined in Planning Analytics. For example, instead of a value of 10 you might see 10.0.
  • Precision, when displaying the unformatted value from Planning Analytics. This is due to floating point values in Planning Analytics.
  • Missing values, due to underfed cubes. Now that MDX is run by Planning Analytics, Cognos Analytics sends NON EMPTY suppression options defined in the query to Planning Analytics. The default Planning Analytics behavior for underfed cubes occurs. This can result in missing values, among other issues with underfed cubes. You can remove suppression from the query. It is recommended that you feed your cells to avoid issues.
  • DivByZero error cells that were returned when dividing a value by null now return null.
  • Invalid members used in a report or dashboard may cause queries to fail, where they were silently removed from the query before. These invalid members are always explained when you validate a report.
  • Member Description property values that were undefined in Planning Analytics are now null instead of defaulting to the Name property. You should use the [<Hierarchy> - Name] property instead, if this was your intent.