Using rbd Migration Tooling in Our Ceph Operations
summary
DigitalOcean runs many Ceph clusters per region and migrates thousands of RBD images using three tools: RBD mirror via Gantry for unattached images, QEMU block jobs orchestrated by Elvos to copy live-attached volumes with convergence and an archived source kept for hours, and Seale to auto-propose batched moves when cluster fill is uneven. Migrations add measurable latency today; approvals and job launches happen with one-click Slack approvals, and ongoing work targets lower IO impact and cross-tier migration.
AI-written summary · about
chapters (7)
- 0:00Introduction and team
- 1:50Reasons for multiple Ceph clusters
- 3:57Gantry migration tool
- 5:41Elvos live migration
- 7:30Seale automation tool
- 9:14Tool comparison and future plans
- 10:18Q&A on latency
related talks
38mGeographical Redundancy with rbd-mirror: Best...
40mJuggling Petabytes: Managing Ceph at Scale with Ceph-ansible
36mRook: Intro and Deep Dive With Ceph Storage
29mTowards Sustainable Multi-Cluster Management in Real-World Academic Infrastructure
44mContainer Management at Scale
36m200 Clusters vs 1 Admin
28mA Welcoming Ceph Experience Improvements for First-Time Users or: How to Make Beginners Love Ceph
50mOps issues