Fellow ranchers, I have a typical situation with some processing within my system which is a cause of some grief. We have a batch job engine (a session bean) which is responsible for invoking various methods from various beans which have been classified as a 'batch' job during design time. This batch job engine calls the respective methods from the other beans. Each is called in the context of a new transaction. Some of these methods are unfortunately rather intensive in their processing and could take a while to complete. This fact is clashing with the container settings for transaction timeouts and causing rollbacks before the method has a chance to complete. We can ramp up the timeout... but this would impact all the other parts of the system which more or less are running happily. We have tried, whereever possible, to design the jobs such that they perform discrete bits in a logical ordered fashion so that we do not experience the timeout situation... but in some instances this presents a rather fragmented approach to finishing a job and in some cases does not give any benefit.. especially in some load situations.. I am sure this is a problem faced atleast by some of you.. any ideas on how to tackle these? any ideas.. be it design/code structure/container settings.. i am ready for all.. :-)... eagerly awaiting ur replies.