Workload Distribution / Parallel Execution in JAVA -

I have a situation here where I need to work in many Java processes running in various JVVs, possibly due to various machines .

Let's say I have a table with a record of 1 to 1000. I'm looking to collect and deliver work 10 sets Then 11 to 200 record to record the worker three, etc. The worker does not have any unnecessary work to say unless the worker can do two things.

This example was purely based on the database, but can be expanded in any system, I believe this file processing, email processing and even further.

I have a small feeling that the immediate reaction will be for the master / worker approach. However here we are talking about various JVMs. Even if a JVM came down, the other JVM should do its job.

Now there are millions of dollars questions: Is there a good framework (production is ready) which will allow me to do this. Even if there are solid implementations of particular needs such as database records, file processing, email processing and their choice.

I have seen the Java parallel execution framework, but I'm not sure that it can be used for different JVMs and if someone has to come down then others will continue to run. I believe workers can be on many JVMs, but what about the master?

More info 1: Hadadop will be a problem due to the need for ZK 1.6. Thats bit too much.

Thank you, Franklin

You can also use message queues. There is a process that creates a list of work and package it in good little ones. It then plots those affair on a queue, waits for workers to appear something on each line. When this happens, then the worker draws a part of the queue band and processes it if a process goes down, then some other process will become sluggish, and people are doing it for a long time, so a lot about this on the net There is some information.