If you intend to use one consumer, run it at the end of the pipeline in a separate thread behind the output queue. It will be a sink for all the pipelines you deploy. Making CC run as part of the processing pipeline has performance implications since it does expensive IO. Any reason why you would not want to run your CC in a separate thread behind the OutputQueue?
The size of the input and output queues should be at least the size of the CAS Pool. Jerry
