The parameters for the Ace Instances (PSL Writers) can be changed on the preferences page of the engine. This document describes the configurations required to control the periodic calculation of results.
The PSL Writer is run periodically. How can we reconfigure the period it runs on?
The parameters for the PSL Writers can be changed on the preferences page of the engine. There are two configuration parameters that influence the length of the cycle:
Calculate service level every x minutes - This parameter contains the maximal length of the engine cycle. When the cycle exceeds this time it is restarted. When the engine finishes everything it has to do during the cycle before this time ends, it can either start a new cycle or it can wait until the end of that time. The exact behaviour is defined by the "Sleep on Idle Time" parameter.
Sleep On Idle Time - This parameter determines what to do when the engine finishes its work before the cycle end time is reached. The values are different in the UI and in the DB. In the DB the two values are “Yes” and “No” – “No” means start the next cycle immediately and “Yes” means wait till the timeout is reached. In the UI the two values are “Continue” and “Wait” – Continue” means start the next cycle immediately.
By default, the cycle time is 1 day. Setting the cycle length to a small value might harm performance since at the beginning of the cycle there are certain initializations and preparations that take time. This time does not depend on the cycle length. Thus the shorter is the cycle, less time is left for actual calculations during the cycle. For example – if the initialization takes ten minutes and the cycle length is set to one hour, there are 50 minutes spent on actual calculation during the hour. If the cycle length is set to twenty minutes, there are three cycles during a hour but 30 minutes from those cycles are spent on initializations and only 30 minutes on actual calculations.
If you require multiple ACE instances in order to improve performance, we do not recommend more than one per CPU core. Instructions for doing so are in the DocOps pages: