Uploaded June 2026 | Updated September 2026, 2 weeks ago
Slides: ceph.io/assets/pdfs/tech-talks/dual_layer_durability_xinnor.pdf
Three-site Ceph with replica=3 is considered best practice, but when a site goes offline, your effective replication factor silently drops to 2, and a single additional disk failure can halt I/O entirely.
This talk addresses that gap head-on. We'll show how combining Xinnor xiRAID with OSNexus QuantaStor 6.7 adds a transparent RAID protection layer beneath Ceph's OSDs, absorbing multiple simultaneous drive failures at the node level without Ceph ever seeing a problem. The result: multi-site clusters that hold their durability guarantees even under the compound failure scenarios that large-scale deployments inevitably encounter. We'll cover the architecture, walk through real failure scenarios, and demonstrate how QuantaStor's native xiRAID integration keeps operational complexity firmly in check. Speakers will be Steven Umbehocker, CEO of OSNexus, and Dmitry Livshits, CEO of Xinnor.
Slides: ceph.io/assets/pdfs/tech-talks/dual_layer_durability_xinnor.pdf
Three-site Ceph with replica=3 is considered best practice, but when a site goes offline, your effective replication factor silently drops to 2, and a single additional disk failure can halt I/O entirely.
This talk addresses that gap head-on. We'll show how combining Xinnor xiRAID with OSNexus QuantaStor 6.7 adds a transparent RAID protection layer beneath Ceph's OSDs, absorbing multiple simultaneous drive failures at the node level without Ceph ever seeing a problem. The result: multi-site clusters that hold their durability guarantees even under the compound failure scenarios that large-scale deployments inevitably encounter. We'll cover the architecture, walk through real failure scenarios, and demonstrate how QuantaStor's native xiRAID integration keeps operational complexity firmly in check. Speakers will be Steven Umbehocker, CEO of OSNexus, and Dmitry Livshits, CEO of Xinnor.
