Re: [rules-users] Shared Facts across nodes

2013-05-22 Thread rtanikella
Hi, I have similar requirements as Mauro in that we want all facts to be shared across multiple drools servers in the cluster. Does anybody have this working with Infinispan or any other framework? Since Infinispan is provided as a service in JBoss AS 7, are there any plans for drools server to

Re: [rules-users] Shared Facts across nodes

2013-04-09 Thread mauro
Hi, as an alternative, would it be possible to have several independent nodes persisting their sessions into a remote and shared database, thus a third listener application would load the database content in memory as an entire session, combining all persisted sessions? Is there an out-of-the

Re: [rules-users] Shared Facts across nodes

2013-04-09 Thread Stephen Masters
Not as a realtime working memory. And don't even think about persisting these things to a database if you're trying to process high volumes of events with minimal latency. I'm assuming that you're trying to process high volumes of events, and you're experiencing performance problems, because

Re: [rules-users] Shared Facts across nodes

2013-04-09 Thread mauro
Steve, the system is not implemented yet, but that's what I'm worried about, you're right. I'll take a look at CEP and distributed queues.. haven't thought of that before. Moreover, does Drools provide any failover solution? Thanks for the light!! Mauro -- View this message in context:

Re: [rules-users] Shared Facts across nodes

2013-04-09 Thread Stephen Masters
With regards failover, there is session persistence, but you won't have a 'hot' failover node to switch to there. However, if your primary servers fail for some reason, you could bring up another instance to take over that session. There may be issues if you have a CEP environment with realtime

Re: [rules-users] Shared Facts across nodes

2013-04-09 Thread mauro
Things are clearer now! I'll analyze the environment and try to apply one of the solutions. Probably I'll be back with some more questions later. Thank you!! Mauro -- View this message in context: http://drools.46999.n3.nabble.com/Shared-Facts-across-nodes-tp4023210p4023267.html Sent from

Re: [rules-users] Shared Facts across nodes

2013-04-05 Thread mauro
Hi guys, I appreciate your promptness!! @salaboy, thank you for the clarification. I'll skip Grid then.. however replication is not exactly what I need. @laune, at the first boot the server will be empty. I don't need initial (static) facts to be loaded, though it has to keep facts inserted over

Re: [rules-users] Shared Facts across nodes

2013-04-05 Thread Wolfgang Laun
On 05/04/2013, mauro mdurant...@gmail.com wrote: One thing I couldn't figure out is how to trigger events (fusion) based on facts contained on different servers. Since servers won't contain the same facts, how to listen to N servers? To clarify, an example: Suppose there's a rule which

Re: [rules-users] Shared Facts across nodes

2013-04-05 Thread mauro
So if I want to know the whole world I have to scale up, right? Ok, I'll try to work with this.. Anyway, is there any plan to implement shared sessions for further releases? Thanks, Mauro -- View this message in context:

[rules-users] Shared Facts across nodes

2013-04-04 Thread mauro
Hello all, After searching the web and Drools forums for a couple of days, I couldn't find a solution to the problem I've got to solve. Anyway, if there's a thread pointing at the same problem, I apologize and please, let me know where it is. My current scenario involves a Drools cluster and

Re: [rules-users] Shared Facts across nodes

2013-04-04 Thread Mauricio Salatino
Hi Mauro, At this moment you can't share the session state between multiple nodes. You can replicate your sessions in different nodes. Which means that two or more nodes will contain the same set of facts. Drools Grid was an experiment, so I don't recommend you going on that direction. On

Re: [rules-users] Shared Facts across nodes

2013-04-04 Thread Wolfgang Laun
Processing a fact does take some time. But there's no guarantee that distributing whatever results from this processing to another node and incorporating it there into an application results in any considerable saving as compared to distributing the original facts and repeat the processing