Ceph Days London 2026

A look at the Ceph roadmap, current development priorities, and the latest activity in the Ceph community.

Our annual Ceph storage roadmap session starts with everything that is happening in the upstream project this year and what we have planned for the future, and closes with the state of what is backed by vendor-supported products at IBM.

Architectural familiarity with Ceph is required. This session contains zero vendor pitches, and it is a caffeinated tour of what the Ceph community is working on at the feature level.

Hang on to your hats, and bring questions!

Organizations are increasingly moving away from traditional VMware‑plus‑SAN architectures and adopting open‑source alternatives such as Proxmox VE with Ceph. In this talk, I share our real‑world migration journey, covering not only the move from VMware/SAN to a Proxmox/Ceph platform but also our transition from a VM‑based network file‑sharing cluster (OpenAFS/NFS) to CephFS backed by SMB. I will discuss the technical challenges we encountered, the architectural decisions behind our new virtualization and storage stack, and the strategies we used to move production workloads with minimal disruption.

Pool migration is a new feature being developed to allow all the objects in a pool to be migrated to a new pool non-disruptively, to facilitate changing the erasure code profile and converting between replica and erasure coded pools. We will discuss how the this feature works, what limitations there will be, and some of the challenges we found during development.

Lee has been working on Ceph for over 2 years, this lightning talk is sharing his insight and experience with teuthology and the optimisations that will reduce execution time of your tests and improve the efficiency of systems within the teuthology suite which will speed up execution times of the lives of developers within the Ceph community for developing fixes.

The Tentacle release of Ceph introduces key NVMEoF enhancements, in this lightning talk we summarise the new features:

High Availability (HA) and Performance:

• Improved Failover Logic: Tentacle overhauled the NVMeoF Gateway fast-failover logic. Beacon timeouts and improve monitoring of gateway health.
• Auto listeners
• Improved Load Balancing: Enhanced automatic namespace load balancing across available gateways to improve performance and resource utilization.

Enhanced Management and Diagnostics:

• New management tools

Dashboard Upgrades:

• The dashboard now support NVMe/TCP gateway groups, multiple namespaces and enhanced multi-cluster management

As well as many more features.

This session explores the architectural shifts required to allow clients to bypass the Primary entirely and fetch data directly from the OSDs. We will discuss the mechanics of offloading data retrieval, decoding, and validation to the client side. Finally, we will share an in-depth performance analysis, highlighting the reduction in overall network latency and improvements in tail latencies across various workload profiles.

Modern frontend development ecosystems, particularly Node.js, are notorious for generating enormous metadata footprints. A single 'node_modules' directory can easily contain over 100,000 files and deeply nested directory trees. Now multiply that by dozens or hundreds of concurrent developer workspaces, all backed by CephFS, and you have a perfect storm hitting your MDS layer.

In this talk, we share our real-world experience delivering on-demand developer workspaces to engineering teams using the DevWorkspace Operator on Kubernetes, with Rook-Ceph providing persistent storage via CephFS. Our primary challenge was workspace startup time: when a developer launches their environment, dependency resolution and filesystem population generate a massive burst of metadata operations: 'open', 'stat', 'readdir', 'create', that can overwhelm a single MDS daemon and turn a quick "start my workspace" into minutes of waiting.

We will walk through the journey from identifying the MDS as the bottleneck to implementing a multi-layered tuning strategy on Ceph Squid (v19) that dramatically improved the developer experience

This talk will begin with a brief introduction of the Ceph-SMB service followed by an overview of the latest developments in the project. We will highlight newly added features such as I/O throughput and rate limiting, Bring Your Own Key (BYOK) support, and enhanced share-level metrics. Finally, we will provide a sneak peek into upcoming features planned for future releases, offering insights into the evolving road map of Ceph-SMB.

AI tools, specifically Large Language Models (LLMs), are increasingly being used to write code, review changes, and assist with operations, under the promise of significant productivity improvements (up to 50%). This raises a practical question for the Ceph community: what happens when AI becomes a contributor?

This talk focuses on real-world ways LLMs can and, in fact, does support Ceph development and operations today.

The Crimson project continues from strength to strength. In this talk we provide a concise update on the latest capabilities for the project.

Managing TLS certificates in a distributed Ceph cluster can be complex and error-prone, with expired or misconfigured certificates often leading to service disruptions and operational overhead.

This lightning talk introduces the certmgr module in Ceph’s cephadm, which automates the lifecycle of TLS certificates across the cluster. We’ll briefly explore how certmgr handles certificate generation, distribution, and renewal, helping operators maintain secure communication with minimal manual intervention.

When a benchmark finishes, it gives you a number. Active benchmarking asks: why this number and not more?

The Ceph Ambassador Program is excited to recognize outstanding contributors who've driven the Ceph community forward through leadership, innovation, and teamwork. These awards celebrate contributor achievements in development, advocacy, documentation, and user adoption, and shine a light on organizations that embody the open source spirit and boost the Ceph ecosystem.

This talk describes the physical relocation of a live Ceph cluster between datacentre rooms, with minimal service interruption, and with bonus hardware lifecycle replacement works along the way.

In this talk, we will present Ceph Dedup for RGW objects, a scale-out deduplication feature that runs on RGW servers and efficiently identifies and eliminates duplicate full RGW objects. The feature is available in the main upstream code today and is targeted for the Tentacle release as a tech-preview.