Parallelism inside nested Transactions

This section is intended for discussion about the usefulness and expectations derived from parallelism inside transactions. The beginning for multiple nesting levels of parallelism is already done, running garbage collection or other asynchronous tasks in parallel with the computation. Another successful example is the asynchronous checking for aborted transactions [here].

Where are the limits for adding new levels of parallelism and what performance can we expect from emerging approaches? Amongst others these are the basic questions we would like to discuss here.