Ceph Days Zurich

A look at how AI is shaping the future of Ceph and what changes lie ahead for the project.

Real-world recovery stories, user impact, and storage architecture adaptations after weeks-long downtime at the University of Zurich.

A practical side-by-side comparison of day-2 operations, architecture, and trade-offs without requiring prior Kubernetes knowledge.

Practical workflows and playbooks with cephadm-ansible to handle host preparation before the orchestrator takes over.

Handson war stories and production pitfalls you should avoid when running Ceph at scale.

Concrete methods and architecture choices to optimize data footprint and curb hardware costs in Ceph.

War stories on repurposing older hardware to hedge against soaring RAM and NAND prices.

How Infomaniak automated per-OSD forensics and disk remediation at scale using their Go-based tool Ceph Companion.

Methodology and findings from Viridien’s ongoing 300PB+ migration from POSIX clusters to Ceph RGW.

Benchmark results (using elbencho & warp) evaluating the latest Fast Erasure Coding optimizations in Ceph RGW

Design and multi-MDS coordination of a single-command kill switch to isolate compromised workloads during ransomware attacks.

How CERN modified restic to use ceph.dir.rctime extended attributes, drastically reducing MDS load during daily backups.

Using directed multigraph algorithms to reduce upmap proliferation by >80% on petabyte-scale production clusters.

An introduction to Rook's automated OSD replacement model and how GitOps handles imperative storage operations.

Introducing rgwtrace, a low-overhead eBPF tracing tool that maps S3 requests down to RADOS pools, PGs, and OSD sets.

Internal data structures, edge-case behaviors, extended monitoring, and throughput-tuning for RGW deletion queues.