Beyond Balancing: How Upmaps Were Used To Avoid Disaster! - Bryan Stillwell, Akamai Technologies @Cephstorage
Beyond Balancing: How Upmaps Were Used To Avoid Disaster! - Bryan Stillwell, Akamai Technologies  @Cephstorage
Uploaded November 2025 | Updated September 2026, 2 weeks ago
Beyond Balancing: How Upmaps Were Used To Avoid Disaster! - Bryan Stillwell, Akamai Technologies

Upmaps are great for nudging a few PGs around to distribute data evenly across your Ceph cluster, but what if I told you they could also save your cluster from a meltdown caused by a sleeping hardware issue that only shows up after a reboot?

In this talk, I’ll walk through a real-world example where a hardware flaw across thousands of Ceph servers (spanning multiple clusters) created a ticking time bomb. I’ll show how we used upmaps to avert disaster and keep things running smoothly. I’ll also share other clever (and maybe unexpected) uses for upmaps that could make your life as a Ceph admin a little easier, and a lot less terrifying!
Beyond Balancing: How Upmaps Were Used To Avoid Disaster! - Bryan Stillwell, Akamai TechnologiesSMB Keeping - Clients, Clusters & Clouds - John Mulligan, IBMIntroduction & State of the CephalopodCeph Developer Summit  - NVMe oF gatewayKeynote: Building Together: Open Storage for Today and Tomorrow - Joachim Kraftmayer, CEO, CLYSONVMeoF in Ceph - Whats New and Performance Best Practices - Mike Burkhart & Aviv Caro, IBMCeph CI and Developer Tooling: Past, Present, and FuturFastEC in Ceph Tentacle A Technical Deep Dive into Next Gen Erasure Coding PerformanceCephalocon 2025 Highlights – Vancouver’s Biggest Ceph Community Gathering!Ceph User + Dev Monthly Meetup - May 2026BXC - An Architecture for Lazy RBD Imaging and Snapshot ExportsKeynote - State of Ceph
Ceph |

Beyond Balancing: How Upmaps Were Used To Avoid Disaster! - Bryan Stillwell, Akamai Technologies

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER