Solving mulitple datasets in parallel

[Amarjith Vishnu]

Hi,

I will be solving multiple datasets in parallel with optaplanner. Solving will be triggered by different end users so count of parallel datasets being run is not known to me.

Will my system break if too many problems are sent to optaplanner?

Should I implement a queuing framework to tackle this?

[Radovan Synek]

Hello,

in general, you can use the SolverManager [1] to solve data sets in parallel utilizing multiple cores of the CPU.

However, if too many data sets are sent, they might not fit into memory, resulting in a JVM crash with OOME. In other words, yes, some kind of queuing is required. There is a quickstart showing integration with ActiveMQ Artemis [2], but it solves only a single dataset at any moment, so you would have to change it to use the SolverManager.

There is another quickstart showing parallel solving on OpenShift [3], which also integrates with ActiveMQ Artemis; just be aware that the optaplanner-operator the quickstart relies upon is still rather experimental.

[1] https://www.optaplanner.org/docs/optaplanner/latest/planner-configuration/planner-configuration.html#solverManager

[2] https://github.com/kiegroup/optaplanner-quickstarts/tree/stable/technology/java-activemq-quarkus

[3] https://github.com/kiegroup/optaplanner-quickstarts/tree/stable/technology/kubernetes

Best regards,

Radek

--
You received this message because you are subscribed to the Google Groups "OptaPlanner development" group.
To unsubscribe from this group and stop receiving emails from it, send an email to optaplanner-d...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/optaplanner-dev/4e61b2f7-b71f-4e7f-8eca-cf51b216f7a3n%40googlegroups.com.

--

Radovan Synek

Principal Software Engineer, Business Automation

Red Hat EMEA

Brno, Czech Republic