Hi!
Lennart Petersson wrote:
> Using TX_SUPPORTS on both the session bean method and the entity bean
> getter methods so unless the container performs the request in some
> other tx, no tx will be used. But even so, yes the request is short.
Good. That is an essential part to any scalable transactional system.
See recent EJB-INTEREST postings by Ian McAllion for more info on this
topic.
> But
> still - sometime we maybe hit the limit where there are that many client
> that are calling the same entity bean instance at the same time that
> they will realize a smaller delay - or?
You should already know the answer to this question by now.
I iterate: instance management in JBoss done with the
one-instance-per-pk method. Calls to a particular entity (=pk=instance)
are thus serialized on a transaction basis, i.e. even if there is no
thread using an instance another transaction must wait until the current
transaction ends (in your case there are no transactions running, so it
is minimzed to waiting for the actual usage to end).
If actual action performed takes long time -> longer delay
If action performed is short, but many concurrent callers -> longer
delay
This should help you determine if a particular case will incur a long
delay due to congestion in usage.
Note that if the action is simply to retrieve data from a previously
loaded entity, then the time taken to perform the action will be
extremely small.
I hope this is clear enough to be of use.
regards,
Rickard
--
--------------------------------------------------------------
To subscribe: [EMAIL PROTECTED]
To unsubscribe: [EMAIL PROTECTED]
List Help?: [EMAIL PROTECTED]
