One of the less obvious issues with looking at the TotalBatchProcessingTimeInMilliseconds is that the value can easily be influenced by the internal, which can drive the commit-duration up while we wait for space to become available in the queue.

We should expose the TotalCommitTimeInMilliseconds so that we can determine exactly how much time is being spent in this loop, to determine if there is really a need to decouple the reader from the dispatcher.