WiredTiger introduced the ability to use multiple threads to write back pages for checkpoint in WT-17178. That is only enabled in certain places where it is known to be critical that checkpoints complete quickly.

We should assess whether there are workloads in ASC and Atlas that would benefit from concurrent checkpoints, as well as workloads that would degrade. This project should focus on automatically enabling concurrent checkpoints for workloads that will definitely benefit, and provide guidance about when users should use or avoid using concurrent checkpoints.