thx for your Feedback
Using a single instance of entity for all clients does not work in a clustered environment. You will need to perist data so that other instances in another server of the cluster see the data. Are you using a database?
Hmm, i hoped that in-memory replication would be able to do this job.
Also im not sure if i ever need a true "cluster".
I just need some kind of mirror being able to act as failover server. I dont need loadbalancing - im only interested into avoiding downtime. (Is this already a cluster?)
So something like
"master" => "read only slave till failover"
would be more then enough.
I thought that modern application servers might have the ability, to cluster beans in-memory within their transaction scope. Yeah i know, that this will lead to the problem, that a normal shutdown will throw avay all instances, but currently - after shutdown - i dont need them any longer (they got a short-time life cycle).
Maybe i just try to archive the wrong goal. The idea was to avoid a locally installed database on our appserver to reduce maintenance efforts. (But maybe it will be already enough to use an embedded db like derby? Seems like some appservers deliver an embedded db to store session states.)
Or is it even worse and both nodes need to be connected to the same db instance?
Seems like there are still huge gaps inside my knowledge ;P
thx again
Bernhard