For many system administrators, Cinder's thin provisioning has been the source of great pain, from deciding the Cinder configuration to figuring out scheduler failures. Since Cinder started supporting thin provisioning, administrators have been facing challenges related to wrongly implemented backends, bugs in the core implementation or just simply a lack of proper understanding of the feature.
This miracle presentation will deep dive into the Cinder scheduling process for thin provisioned volumes, covering existing functionality and the latest improvements as well as those yet to come. By the time we reach the conclusion your thin provisioning headaches will be miraculously cured, or at least the pain will have been relieved, guaranteed.
Understand the scheduling of thin volumes.
Be aware and learn the common mistakes admins can do.
Understand what the new auto_max_oversubscription_ratio means and how that can help in the daily basis.