Hi, I'm currently evaluating Drools 3.0 and trying to understand how to build an architecture that would meet the following requirements:
a. ~2000-4000 rules b. ~20 events per second c. Stateful execution d. On the average, large number of events (~100,000-200,000). e. 24x7 execution f. Ability to recover from failures (will require persisting both the working memory events and the RETE network status). * Has anyone tried running Drools with these kind of requirements? * Has Drools even been tested with 100,000 objects? What is the memory signature of the RETE network in this case (let's say that each object size is 3KB). * I've seen that the "near future" roadmap of Drools including working on events that are not in the working memory. Could you provide more details on what is means? (how close is "near future" by the way?) * Can the rule engine be persisted (is it serializable)? * Would you recommend on running the events in semi-bach mode (injecting a chunk of events to the engine, fire rules, persist the engine and inject another chunk of events)? Thanks, Avi ____________________________ Avi Kavas, SOA Center, [EMAIL PROTECTED] direct: 972-3-5399443 mobile 054-4497198 MERCURY www.mercury.com ______________________________________________________________________ This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email ______________________________________________________________________
