Hi, I am new to Clara and trying it in one the projects. Has anyone tried to use Clara with Hazelcast as its working memory? In general, any tips on the WorkingMemory management for an enterprise app with many domain objects (50 types, each about 100k objects) would be really helpful.

Btw, I hope this is the right forum to ask Clara related questions.

@ramkumarkb: This is indeed the correct place to ask questions about Clara. 🙂 We’ve used Clara in some heavy-hitting scenarios, although perhaps not quite as many as 5 million total objects, at least intentionally. In general, my instinct is that bad scenarios would probably be a combination of both the number of facts, the rules, and the data scenarios, rather than just the number of facts though. That’s generally what we’ve seen, except in some cases where we’ve had bad rules create a number of facts that was a problem for the heap. I’m not sure what you mean by using Hazelcast as the memory - are you implementing your own memory as a replacement for the Clara defaults, considering using it in combination with the persistence stuff @mikerod is working on with https://github.com/rbrush/clara-rules/issues/198, or something else? I’m not an expert on Hazelcast, this is the first time I’ve heard of it. It would certainly be interesting to me to hear about other perf-intensive use-cases.