Cephalocon 2025
| title | author | length |
|---|---|---|
| Opening Ceremony & Land Acknowledgment | Nova Wolf | 10m |
Introduction in a traditional language, sharing a hereditary name and role as a knowledge keeper bestowed at age 25, and a brief account of growing up in a… | ||
| Keynote: Welcome & State of the Cephalopod Part I | Josh Durgin, Neha Ojha, Dan van der Ster | 09m |
Opening address notes this is the first Cephalocon in North America and gives concrete project health: 1,400 contributors, 700K+ lines changed, 156K+ commits… | ||
| Keynote: IBM’s Commitment, Investment, and Direction for Ceph in 2026 | Gilad Sade, Neha Ojha | 13m |
IBM commits Ceph as core storage across at least 15 IBM products, provides nearly 80% of recent contributions, and has doubled the number of paid Ceph… | ||
| Keynote: Building and Sustaining Momentum in Open-Source Storage | Brett Kelly | 14m |
Company experience selling Ceph hardware and running real clusters shows the core problem: device density outpaces cluster recovery and IO plumbing. Tests… | ||
| Keynote: Growing Ceph Knowledge in the User + Developer Community | Laura Flores, Neha Ojha | 11m |
Monthly Jitsy meetups connect users and developers to turn feedback into code and docs, producing 15 action items so far, fixes to documentation, a plan to… | ||
| Keynote: Real Talk: A Ceph Users Panel | Laura Flores, Neha Ojha, Dan van der Ster | 40m |
Six Ceph users describe concrete deployments: block, object and CephFS at cloud providers, universities and labs; one site runs 25 PB legacy + 45 PB new… | ||
| Keynote: Welcome Back & State of the Cephalopod Part II | Josh Durgin, Neha Ojha, Dan van der Ster | 02m |
Opening housekeeping: a code of conduct reminder, the solution showcase continues today, and sponsors listed as 45 Drives Bloomberg IBM. Announces a t‑shirt… | ||
| Keynote: Powered by People: Growing the Ceph Community Through User Engagement | Anthony Middleton | 09m |
Since May the Ceph Foundation reports concrete community gains: 12 blog posts published, 700 new LinkedIn followers, and plans for about six Ceph Days already… | ||
| Keynote: From Busy to Impactful: Maximizing Ceph’s Collective Resources | Joseph Mundackal | 11m |
Bloomberg describes Ceph as production storage that runs at enterprise scale — ~1 million virtual CPUs backed by few hundred PB raw storage and georeplicated… | ||
| Keynote: Building Together: Open Storage for Today and Tomorrow | Joachim Kraftmayer | 04m |
Company increases financial support to the Ceph Foundation to its highest tier this month and reports concrete engineering work: performance tuning, cluster… | ||
| Keynote: Ceph Foundation Update: The Good, the Bad, and the Ugly | Matthew Leonard | 14m |
Ceph Foundation reports concrete changes: 26 members with Western Digital and Samsung rejoining and Sony newly on board; a paid community manager (Anthony)… | ||
| Keynote: Ceph Developers AMA: Meet the Minds Behind Open Source Ceph | Daniel Gyovich, Matt Benjamin | 21m |
Ubuntu 24.04 lacks a ready Ceph package image so installs often fall back to 22.04 or container deployment; the project is shipping upstream containers while… | ||
| Keynote: Tentacle Awards | Federico Lucifredi, Gaurav Sitlani | 14m |
Cephalocon honors contributor work with concrete wins: Tentacle Awards 2025 go to Casey Bodley for steering RGW through major feature backports while balancing… | ||
| MSR(Mutli-Step Retry): An Generalization of CRUSH Allowing Multiple OSDs Per Failure Domain | S. Just | 24m |
Claim: MSR generalizes CRUSH so a placement rule can allow multiple OSDs per failure domain and still keep deterministic, weight‑proportional, and stable… | ||
| Delivering Enterprise-Grade Object Storage: Bloomberg’s IAM Journey in a Multi-Tenant Arch... | S. Gao | 38m |
Ceph-based clusters handle 2 billion requests/day and 250 TB/day across 10 clusters (largest at ~4,500 OSDs per site) backing ~3,000 tenants and ~77 PB total… | ||
| Enterprise-Ready Ceph: Simplifying Storage Management, Operations, and Deployment | Steven Umbehocker | 20m |
OSNexus presents QuantaStor as a turnkey platform that installs from an ISO, turns bare metal into managed Ceph appliances and exposes every feature via REST… | ||
| Cluster-scale Problem Solving | Kellen Renshaw | 10m |
Canonical Support Engineering uses Ceph to turn vague goals into measurable SLAs: they qualify and quantify workloads (example target: 100,000 IOPS), group by… | ||
| Using rbd Migration Tooling in Our Ceph Operations | Daniel Radjenovic | 12m |
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… | ||
| Ordered Bucket Index Shards in RGW | Eric Ivancich | 26m |
RGW moves from hash-based shards to maintaining lexical ordering across shards so listings iterate shards sequentially and avoid querying every shard; shard… | ||
| Could Ceph Become More Like a Database? | Radoslaw Zarzynski | 48m |
Proposes adding RADOS Native Indexing primitives so an object write can carry a tag that updates its index atomically at the OSD in a single RADOS transaction… | ||
| Ceph Object Storage Roadmap Session at Cephalocon 2025 | Daniel Parkes, Matthew Benjamin | 36m |
Squid adds object headers with replication status for workflow checks, per-bucket S3 logging with a synchronous journal mode for incremental backups, RGW IAM… | ||
| Detect, Decide, Fence: Resolving Netsplits in 3-AZ Stretch Cluster | Kamoltat (Junior) Sirivadhna | 27m |
Concrete netsplit resolution for 3‑AZ Ceph: monitors publish connection scores to build a location graph, the system runs Bron–Kerbosch to find maximal cliques… | ||
| What's (in) a Build System? | Zack Cerza, John Mulligan | 37m |
Ceph's build system moves from brittle, multi-script Jenkins jobs to containerized, reproducible builds using the build-with-container (BWC) tool and new… | ||
| Inking Out Inefficiencies in Ceph Erasure Coding | Alex Ainscow, Bill Scales | 37m |
The Tentacle release ships a new Erasure Coding implementation called fast EC that boosts small‑read IOPS roughly threefold versus legacy EC and reduces space… | ||
| Ceph RGW Bucket Snapshots: Beyond Versioning | Yehuda Sadeh-Weinraub | 32m |
Ceph RGW adds bucket-level snapshots by extending S3 versioning with a monotonic snap ID dimension so every object instance maps to a snapshot and listings can… | ||
| Rotating Heads and CephX Keys | Patrick Donnelly, Yehuda Sadeh-Weinraub | 29m |
CephX moves its default rotating-service cipher from AES‑128 to AES-256k, introduces an o rotate API to atomically update credentials, and migrates the monitor… | ||
| CBS: A Containerized, Declarative, Flexible Build System Enhanced for Ceph Builds | Joao Eduardo Luis | 30m |
CBS is a build system that assembles custom Ceph releases by producing signed RPMs, pushing them to S3, and composing final container images pushed to a… | ||
| A Harbor-side Chat on CephFS | Patrick Donnelly | 36m |
Tentacle adds case-insensitive subtrees (per-directory, Unicode-normalization and encoding choices) and a file block diff API that already landed in code and… | ||
| Fragmentation in Ceph: Why It Happens and How We Fix It | Mark Nelson | 29m |
Fragmentation is a primary root cause of decade‑long Ceph performance regressions, arising in FileStore on XFS and Btrfs from inode splits, allocation group… | ||
| How Simple Could Ceph Be? | Ernesto Puerta | 32m |
Ceph’s complexity traces to structure, function and operator cognitive load: the codebase exposes 2057 config options while only ~70 basic options are… | ||
| Demystifying Hyper-Dense Ceph: How To Design for Large Capacity NVMe | A. Goncalves, M. Nelson | 31m |
Claims that hyper‑dense Ceph with >30TB NVMe can match availability if layout and device settings match real device characteristics, supported by concrete… | ||
| CephFS Tell-Me-What-Happened --Yes-I-Really-Really-Mean-It | Anthony D'Atri | 08m |
Provides a persistent, queryable audit of CephFS and cluster commands backed by LibCephSQLite, recording every ceph fs subvolume rm, ceph config set and other… | ||
| Enhancing Ceph for Infinite Scalability at LINE | Jinmyeong Lee, Sungjoon Koh | 33m |
LINE builds a multi-backend object service that fronts multiple Ceph subclusters and a cost‑efficient S3 store (Dragon) behind a single endpoint, operating 20+… | ||
| Enhancing Ceph Monitoring at CERN: Our Approach and Solutions | R. Valverde Cameselle, E. Bocchi | 38m |
CERN migrates monitoring from Graphite to Prometheus + Thanos since 2018, storing five years of metrics with 15s raw resolution and automated downsampling on a… | ||
| Evaluating the Performance and Power Efficiency of Ceph Storage Configurations | Tom Byrne | 34m |
Benchmarks compare three 24‑drive node types (8 TB SATA, 22 TB dense HDD, 15 TB TLC NVMe) using RADOS benchmarking across EC layouts (notably 4+2 EC and 8+3… | ||
| Deploying and Operating CephFS for Scientific Applications at Fermilab | Alison Peisker | 30m |
Fermilab runs CephFS in production since spring 2024 after a spring 2023 rollout, using cephadm to bootstrap and manage a cluster with about 1.5 PB raw / 500+… | ||
| Beyond Balancing: How Upmaps Were Used To Avoid Disaster! | Bryan Stillwell | 19m |
Thousands of NVMe servers show a kernel-boot fault: specific rettimer cards in slot groups die from an overvoltage at reboot, risking many PGs losing multiple… | ||
| An Open Source Distributed Rate-limiting Service for Ceph-object and Other HTTP APIs | Jacques Heunis | 15m |
A compact open-source system that enforces cluster-wide request and bandwidth limits for S3 and other HTTP APIs, built around HAProxy + Lua hooks that emit… | ||
| Using Rook in Production on National Research Platform | Dmitry Mishin | 19m |
Experience running 8 Ceph clusters under Rook inside one Kubernetes cluster with a multi-petabyte footprint: 13 PB raw / 800 OSDs and about 12,000 CSI volumes… | ||
| Ceph Durability: How Safe Is My Data? | Dan van der Ster | 32m |
A new Monte Carlo simulator models Ceph MTTDL and nines across real configs and confirms some concrete surprises: doubling PGs per OSD drops a triple-replica… | ||
| VLLM K/V Caching With Ceph | Kyle Bader, Tushar Gohad | 44m |
Prototype Ceph K/V caching plugin for vLLM offloads attention states to clustered NVMe and complements GPU HBM, using LM Cache with 256-token… | ||
| Ceph Performance: The Basics of Tentacle Racing | Lee Sanders | 28m |
Argues that disciplined, deterministic benchmarking yields repeatable Ceph performance and shows concrete tooling and practices: CBT YAML workloads add… | ||
| One Port, Many Gateways: Scaling RGW for Uptime & Production Resilience | Jack Shen | 21m |
Bloomberg enables multiple Ceph RGW processes to bind to the same port using the Linux SO_REUSEPORT kernel feature, changes in the RGW startup code (pull… | ||
| Optimizing Scrub Performance: Balancing Scrub Speed Against Client I/O Demands | Ronen Friedman | 40m |
Explains concrete knobs and measured effects: use ceph pg/dump and ceph osd dump scrub reservations plus counter dumps to see scrub reservation queues and new… | ||
| Security Process Improvements in Ceph and Rook | Sage McTaggart, Gabriella Roman | 24m |
Process shifts from manual triage to automated pipelines: scanners (Twistlock, Snyk/MEN, SonarQube, Dependabot) once returned 20,000 false positives, so… | ||
| Guide To Getting Started With Developing in Ceph | Jonathan Bailey, Alex Ainscow | 32m |
Step‑by‑step practical entry: use John Mulligan’s container script that mounts the Ceph source at /sef so local edits build inside a container, increase Podman… | ||
| Doctopus: How Ceph Documentation is Maintained & Evolved, & How You Can Contribute | Anthony D'Atri | 25m |
Ceph docs live in the repo doc/ tree with release-specific branches and a live selector, totaling about 178,000 lines across ~567 files; PRs trigger a full… | ||
| Re-Architecting Ceph, One Crimson-OSD at a Time | Matan Breizman | 28m |
Crimson-OSD is a drop-in replacement for the classical OSD built on Seastar C++ that re-architects the IO path to single-thread-per-core shared-nothing… | ||
| Ceph on Arm64/Ampere: IOPs/Watt Demystified | Federico Lucifredi | 17m |
Comparison of Ampere Altra versus Ampere M1 (MR1) on Ceph shows the MR1 packs more cores per box and delivers higher throughput and efficiency: 400k IOPS at… | ||
| BlueStore, Faster + Smaller = Better | Adam Kupczyk | 33m |
BlueStore gains concrete TCO wins: a write-path transformer now reclaims 10–15% lost space on heavily overwritten RBD images and an alternative scrub-time… | ||
| Asok Gems and How To Polish Them | Marcel Lauhoff | 24m |
Ceph Admin Socket (`asok`) exposes actionable internals and the talk demonstrates practical tooling: a script enumerates 179 commands across daemons, finds… | ||
| Introducing Ceph Mgmt-gateway: Unified Access, OIDC-based SSO, and HA for Manageme... | R. K. Elhichou | 37m |
Ceph adds an nginx-based mgmt-gateway in the Tentacle release that front-ends dashboard, Prometheus, Grafana and Alertmanager, terminates TLS, and enforces… | ||
| Per-Subvolume IO Metrics in CephFS | Igor Golikov, Alex Markuze | 29m |
Per-subvolume IO metrics land inside the MDS using a client-side recorder that tags ops when MDS metadata marks a path as a subvolume, packs each op into a… | ||
| SMB Keeping - Clients, Clusters & Clouds | John Mulligan | 23m |
Ceph Tentacle adds a supported SMB stack on CephFS with Samba containers, a Ceph MGR module, a Prometheus metrics exporter and an SEFS proxy to reduce memory… | ||
| OSD Down! but Why? | Jamie Pryde | 15m |
Step‑by‑step method for a crashed OSD: check the cluster log or run seph‑w / seph health for immediate errors, then open the OSD log, find the abort/stack… | ||
| Scaling RGW D4N Caches With Experimental Offloading | Jesse Williamson, Matthew Benjamin | 18m |
RGW offloads D4N caching to external stores to speed bucket-index work: a single-node Redis prototype exists now and the team is building a FoundationDB-backed… | ||
| From Chaos To Clarity: Systematic Ceph Cluster Analysis for Rapid Troublesho... | S. Zhang, J. Blanch | 27m |
A systematic pipeline ingests Ceph diagnostic bundles, scans releases against a YAML of notorious bugs, and flags imbalanced pools, hot-primary OSDs… | ||
| The Ceph Community: A Driving Force for Project Evolution | Frédéric Nass | 24m |
University project migrates 7,550 TB across 56 labs from SAN/NAS to Ceph because it never loses data (accessibility outages only) and reduces TCO versus… | ||
| How Rook/Ceph Enables Science @ Rubin Observatory | Joshua Hoblitt | 28m |
Rubin Observatory runs a consolidated Rook/Ceph stack that handles ~20 TB of raw images per night, transforms 2–3 GB focal-plane exposures, and pushes data… | ||
| Efficient Ceph Performance Troubleshooting in Production Using eBPF | Dongdong Tao | 38m |
eBPF tools trace every Ceph I/O and produce per-IO lifecycle latencies so one can see messenger, OSD queue, BlueStore substages and acting set at the client… | ||
| Modernizing Ceph Deployments at CERN: CephFS and Object Storage Across Data Centres | Enrico Bocchi | 34m |
CERN runs Ceph at ~100 PBs across ~20 clusters in three locations and reports concrete moves: split huge file systems into multiple filesystems (multifs)… | ||
| Ceph Operations at Scale | Matt Vandermeulen | 29m |
DigitalOcean runs Ceph as primary block and object storage across 75 clusters totaling ~270 PB and ~34,000 OSDs on 2,000 nodes, so automation focuses on… | ||
| Ceph-CSI: The Engine Behind Persistent Storage in Kubernetes | Niels de Vos, Madhu Rajanna | 27m |
Ceph‑CSI provides a full CSI driver for Ceph RBD, CephFS and NFS (three backends) and implements newer CSI features early: volume group snapshots for… | ||
| Ceph Scientific Users Working Group BoF | Dan van der Ster | 39m |
Community-run Ceph Scientific Users BoF convenes a mix of research sites exchanging concrete practices: running RBD with three replicas and CephFS as backends… | ||
| NVMeoF in Ceph - What's New and Performance Best Practices | Mike Burkhart, Aviv Caro | 34m |
Ceph’s Tentacle release ships NVMeoF with multi-gateway Tentacle gateway groups that store config in a shared OMAP and use a monitor with Paxos-style beacons… | ||
| Why We Have 17 Clusters Instead of One (and Why You Probably Shouldn’t Do Th... | O. Kukla, P. Tlapa | 15m |
CDN77 splits storage into 17 clusters (14 HDD, 3 SSD) behind an Nginx load balancer using an nginx chash module that hashes the first four chars of CRC32 from… | ||
| Securing Data in Motion and at Rest: Quantifying End-to-End Encryption Perfo... | V. Umrao, D. Parkes | 29m |
Real-world benchmarks on Ceph RGW quantify costs and show what changes performance: client TLS terminated at RGW has almost no throughput loss and only ~10–20%… | ||
| nothing matches | ||