Making RBD Snapshot-based Mirroring Robust for Disaster Recovery | Ceph Days NYC 2024

The feature to mirror RADOS block device (RBD) images across clusters by asynchronous replication of RBD snapshots was introduced a few years ago. It's been recently integrated into the disaster recovery (DR) solution for container workloads backed by RBD in OpenShift kubernetes environment. The integration and testing of the DR solution uncovered bugs and helped identify missing pieces in snapshot based mirroring. Bug fixes and enhancements have been made to librbd and kernel RBD clients, and in the ceph-mgr service that schedules mirror snapshots. There is still a lot of scope for improvement here. This talk will give an overview of the current state of RBD snapshot based mirroring, its usage in Openshift's Ceph storage disaster recovery solution, and the planned improvements for the feature. Folks interested in using RBD mirroring for their disaster recovery solutions will find this talk especially useful.