Query related to performance in kogito

[Ranjan Das]

Hi Kogito team ,

In one of my kogito quarkus application I am using a jbpm process and calling different rule files from the Business rule task as a rule unit. 

As my object model is hierarchical(has 3-4 level of list objects) ,so to  reduce the iteration at rule level (for enhancing the performance) I am iterating at the process flow level .

But I have seen that even though rule-unit are faster than traditional ruleflow-group of drools but still it creates separate kiebases for each rule-unit.
In such a case , if I have 100 Business rule tasks in my process flow which is calling 100 rule units (one rule-unit from each rule task) it will take a substantial amount of time to load all the kie bases firstly and then once the kiebases are loaded (lets say after 1st call) it might take some time in calling multiple kiebases.

Please let me know if there is any gap in the above understanding.

Also in such a scenario , is there any way so that i can reduce the number of kiebases while retaining the number of business rule task as 100 (so that the orchestration of rules is retained at process flow level). I thought of using agenda group within rule unit  but seems kogito has removed agenda group firstly ,secondly how can i activate a particular agenda group from a jbpm process in kogito ?


Thanks in advance,
Ranjan Das