Run the many-collection-test on several increasingly large datasets, for instance, run the test on collections numbered 5k, 25k, 200k, and so on. Also run the test before and after the changes that were made to reduce stalls during the checkpoints. These test results will be used to write a report as part of another ticket.