Performance is everyone's concern.  Remember that drag race testing in single user mode is not helpful to determine real world experiences of your multi-user application.  On high user count systems with many concurrent READs, PRO/5 users can expect a significant performance boost from BBj's highly efficient algorithms, as can be seen from the chart below. BBj's concurrent READs outperform PRO/5's by up to 10-fold, as the user count grows.  The proof of the pudding is in the eating, and with this new functionality you can move back-and-forth in your testing and evaluate the performance before taking the final plunge. Remember too that PRO/5 batch jobs can execute in BBj directly against the data without any communication protocol overhead, and without code recompilation.