I'm trying to find best way to execute rules from a use case of batch execution. The batch process more than 10 000 rows and need to load a lot of data in memory. The rules should be apply to all rows, i have seen there are some ways to execute rules : directly from rest api kie server after i have deployed a container from workbench for example. Or i could use kie scanner to load new container and execute rules from the application without do a http call. I am worry about performance of kie server rest api, should be a problem? When should we execute rules with rest api vs java application embedded by using kie scanner maven.. ? Drools have so many ways to execute rules.
Asked
Active
Viewed 125 times
0
-
This is too broad. It depends on what you're doing with the rules against your data -- eg your specific use case. I've executed rules in batch against a 1.2-million record dataset, and depending on what i was actually _doing_ with those rules I designed my system (and rules) as appropriate. – Roddy of the Frozen Peas Mar 01 '21 at 15:55
-
Did you use kie server api rest call to execute rules on the 1.2 million record? It should depends of the size of each data, the required externals datas to execute rules, and of course how long the business requirements about how long batch should he takes. – julus Mar 01 '21 at 22:39
-
No it depends on what you're doing. If you're comparing each record to every other and doing lots of iterative work, that's one use case; otherwise if you just run right through once and see what hits, that's a very different. – Roddy of the Frozen Peas Mar 01 '21 at 22:44