Context: You need to export data from an Hyperion Essbase application, through a script, as you need to extract a specific perimeter, and with high performance (especially in case of dynamic calc members to extract).
Objective : Export data with the best possible performance.
Export of essbase data on a specific perimeter, including dynamic calc heavy members, can sometimes be time consuming. Performance becomes then a main focus.
Different design choice have to be considered:
- format of the export:
A first choice has to be done : export in a relational table or in a flat file.
Experience shows best performance with an export in a flat file, compared to an export in a relational database.
2. Options to activate : FIXPARALLEL
FIXPARALLEL will allow performing the calculation (here the dataexport) on different threads in parallel.
Typically, it will have to be activated for a sparse dimension, significantly sparse to allow wide parallelization.
Syntaxe:
The FIXPARALLEL will be used in the same way as a classic FIX, for a single member list of a given dimension.
FIXPARALLEL will also require activation of a number of threads to be activated for the specific calculation.
Ex : to activate 8 threads on the level 0 members of the “total_project” member, we will use the following syntaxe, when fixing the “Project” dimension:
FIXPARALLEL(8,@RELATIVE(“Total_Project”,0))
The fixparallel is closed using the command “ENDFIXPARALLEL”, in the same way as the “ENDFIX” command.
Correct number of threads to be activated is specific to the dimension concerned, and will need to be determined through testing.
Warning : the memory used by the database during export will have to be monitored to evaluate impact. Adjust calculator cache in case of issue.
