14 Appendix A: FSI Optimization Settings
A typical Balance Computation implementation can involve a large number of
computation instances given the number of Data Origins, GL Types and currencies. There
is commonly a bell-curve distribution of processing times for each computation rule.
Some complete in seconds, some complete in minutes, and others may take hours to
complete. The run time of a computation is normally closely related to the volume of
data it generates, however, there may be instances when a computation runs slower than
it must due to stale database statistics, skews in the data being queried, a lack of
appropriate indexes, or a poor execution plan generated by the database optimizer. Your
DBA may be able to enhance the overall performance of the database by gathering database
statistics more frequently, generating additional indexes, or by other tuning
techniques. One such additional tuning technique is to add hints to your queries. The
data from the balance sheet is categorized by the GL types. Within each of these GL
types (Service Tasks), for execution, you can invoke multiple threads in the database.
This process can be governed by the following methods:
- Parallelize Executions using the FSI_OPTIMIZATION_SETTINGS Table
- SQL Hints