I have a Java Application which handles different region for example 10 regions. Each region have different number of tasks.

I encounter a problem is that one particular region (Region A) have lots of tasks and the processing time of each task for that region is very long. Therefore, if I use a Single Thread Pool, and tasks of all region will be submitted to the Single Thread Pool, tasks of Region A will block other region 's tasks. Tasks of other region can only wait in the block queue until the tasks of Region A are finished.

I want to achieve if there are 5000 tasks of Region A submitted to Single Thread Pool , and each task is processing with a long time, other region tasks can be processed in parallel without being blocked by tasks of Region A.

I have a requirement which cannot create too much thread for the Single Thread Pool, and also newCachedThreadPool() is not recommended to use.

What my approach is to create 1 Thread Pool for each region, which will create 10 Thread Pool in total. But I do not know if there are any problems regarding multiple Thread Pool and if there are any other proper way to achieve it.

CodePudding user response：

The processing capacity of running threads is always limited by the number of threads the hardware's CPU provides - keep that in mind when allocating threads to thread pools.

Use newFixedThreadPool(int nThreads) for each region.

If your processing is mostly compute intensive, pick smaller sizes; perhaps newFixedThreadPool(3) for region A and newFixedThreadPool(1) for the others.

Chose larger numbers if your processing spends a lot of time waiting for processing external to the thread (eg web or db calls).

Experiment to see what provides acceptable liveliness.