Delete comment from: MySQL Musings

Mats Kindahl said...

Mark, thank you for your comments.

There is currently not any logic like that in the implementation. Adding a rule to check the length of the "in-progress" queue is easy, and if it is empty, performing an fsync would allow transactions to commit fast when there is little concurrency in the system.

We will perform benchmarks next to ensure that the patch performs as expected, and it is important that the response time is short both for high-concurrency and low-concurrency workloads.

Aug 12, 2011, 7:45:29 AM

Posted to Binlog Group Commit Experiments