Cephalocon 2023
as the organisers put it
View The Schedule Sponsor register now Thank you to all who joined us for Cephalocon 2023! We look forward to seeing you at future events. To experience the best of this year’s event, be sure to watch…
| title | author | length |
|---|---|---|
| Keynote: Welcome & Opening Remarks | Josh Durgin, Dan van der Ster | 05m |
Conference opens with status updates: Ceph moves from Nautilus through Octopus Pacific Quincy to a fifth release (Reef) about to be released, organizers note… | ||
| Keynote: Ceph: The Future of the Storage TODAY | Kyle Bader, Danny Mace | 13m |
Engineers move from Red Hat into IBM on Jan 1 and remain intact, with a firm commitment that Ceph stays 100% open source and upstream first. IBM claims it is… | ||
| Keynote: State of the Cephalopod | Radoslaw Zarzynski, Samuel Just, Yaarit Hatuka, Josh Durgin, Zac Dover | 36m |
Project updates state nearly 1 exabyte reported to opt‑in Telemetry, Reef release shifts to June after large‑scale testing with the Posse supercomputing… | ||
| Keynote: The Road to Open Source Ceph: Bloomberg’s Enterprise Ceph Journey | Matthew Leonard | 18m |
Bloomberg explains moving from in‑house systems to an open source platform built on Ceph and OpenStack, running dozens of clusters worldwide with hundreds of… | ||
| Keynote: Welcome Back & Remarks | Josh Durgin, Dan van der Ster | 02m |
Conference opens with concrete community work: documentation lead Zach Dover reports a day of patches and rewrites aimed at improving docs for beginners with… | ||
| Keynote: Lead Developers Town Hall | Josh Durgin, Samuel Just, Radoslaw Zarzynski, Yaarit Hatuka, Zac Dover | 39m |
Direct answer on upgrades: Ceph supports upgrade two releases so Octopus → Pacific or Quincy works but Octopus → Reef does not. RGW gains faster deletion by… | ||
| Ceph on Windows | Alessandro Pilotti | 35m |
Ports Ceph RBD to Windows by shipping a user-space client (wimbd) plus a custom kernel miniport driver that exposes RBD images as block devices, supports… | ||
| Crimson Project Update | Samuel Just | 32m |
A ground-up rewrite of the ceph-osd targets cutting per‑IO CPU cost by moving from thread-pool handoffs to a user‑space scheduler so a single core handles a… | ||
| Embracing Ceph: Key Factors Driving Companies to Adopt Ceph Storage Solutions | Andras Pataki, Frank Yang, Matthew Leonard, Vasyl Purchel, Luca Cervigni | 43m |
Organizations choose Ceph because it delivers block, file and S3 object from one system, scales to 300 PB/year for SKA-class ingest, and powers a private cloud… | ||
| Using EBPF to Develop a Powerful Tool for Diagnosing Ceph Performance Issues | Dongdong Tao | 39m |
Built an eBPF/uprobe tool that measures per-request Ceph OSD lifecycles across messenger, OSD and BlueStore (receive, dispatch, queue, osd, bstore… | ||
| Ceph Multi-Site at Scale: Bloomberg’s Disaster Recovery Journey | Jane Zhu, Krunal Chheda | 44m |
Bloomberg converts its object store to Ceph multi‑site for server‑side DR, running four clusters (~65 PB raw, ~4,500 OSDs per cluster, >50k buckets, ~1 billion… | ||
| Ceph: Lessons Learned from Emergency Support | Joachim Kraftmayer | 44m |
Emergency support cases across two–three years show concrete fixes and anti-patterns: a three‑OSD cluster is migrated to ~100 OSDs without downtime; RAID… | ||
| Ceph Crash Telemetry - Observability in Action | Yaarit Hatuka | 44m |
Ceph ships a built‑in Telemetry module that 2500 clusters currently opt into, reporting 2500 clusters and nearly 1 exabyte and about 130k OSDs; collection uses… | ||
| Best Practices of Production-Grade Rook/Ceph Cluster | Satoru Takeuchi | 35m |
On-prem Kubernetes runs Rook managing two Ceph clusters: RBD on NVMe + RGW on HDD with NVMe used for indexes and HDD for bulk data, OSDs auto-created and… | ||
| Enhancing Observability in Rook Ceph | Deepika Upadhyay, Gaurav Sitlani | 37m |
Built end-to-end tracing for Ceph runs through Rook using OpenTelemetry instrumentation, a sidecar collector, and Jaeger storage and UI, with plans to replace… | ||
| Ceph Messenger Situation and Optimization | Chunsong Feng | 17m |
Integrates a UCX-based messenger into Ceph to replace the old RDMA path, implementing zero-copy active-message receive and distinct small/large send paths… | ||
| NVMe-Over-Fabrics Support for Ceph | Jonas Pfefferle | 44m |
A NVMe-over-Fabrics Gateway for Ceph translates NVMe/TCP and RDMA to RADOS/RBD using an SPDK target and a Python gRPC control daemon. Configuration persists in… | ||
| The Enterprise Ceph Spectrum | Philip Williams | 05m |
Canonical lays out a Ceph spectrum from single laptops to tens‑or‑hundreds‑petabyte datacenters, using Juju and charms for large deployments, OCI rocks images… | ||
| Kubernetes Stateful Application Disaster Recovery with Zero Data Loss | Daniel Parkes | 46m |
Shows a Kubernetes disaster‑recovery stack that achieves RPO zero by using an external Ceph stretched cluster (replica 4, crash‑map) accessed via Rook CSI… | ||
| Ceph Object Storage Overview, Capabilities and Future Plans | Matt Benjamin | 29m |
rgw is a high‑fidelity S3/Swift HTTP object gateway that scales to hundreds of petabytes in a single cluster and supports bucket namespaces into the hundreds… | ||
| MicroCeph: What, Why and Who Is It for | Peter Sabaini, Utkarsh Bhatt | 17m |
A compact Ceph distribution packaged as a snap package targets small deployments by bundling userland, isolating services, and adding a lightweight management… | ||
| Troubleshooting a Cephadm Deployment | Michel Raabe | 29m |
Cephadm troubleshooting focuses on getting actionable logs not just “scheduled” messages: enable per-service and cluster log files, raise log_level 5 (default… | ||
| Chasing Bad Checksums: A Journey Through Ceph, TCMalloc, and the Linux Kernel | Dan Hill | 20m |
Cluster runs dozens of nodes and hundreds of OSDs serving ~10 PB and sees about five checksum-crash OSDs per week; lab recreate fails every 3–4 days until… | ||
| Rook: Why Would You Ever Deploy Ceph Inside Kubernetes? | Travis Nielsen | 12m |
Rook deploys Ceph inside Kubernetes by implementing an operator + CRDs to express desired state, using the CSI driver for mounts and managing Ceph as the data… | ||
| External Rook Ceph Cluster | Parth Arora, Subham Kumar Rai | 13m |
Describes an external Rook operator pattern that connects a standalone Ceph cluster to Kubernetes via CSI by running a Python fetch script that outputs JSON… | ||
| Advancing Infrastructure with Ceph Technology | Joachim Kraftmayer, Dan van der Ster | 16m |
Ceph is positioned as a scalable, flexible building block with rock‑solid RADOS and CRUSH, but usability and day‑to‑day ops lag; the company offers proactive… | ||
| Multi Cluster Ceph at Scale | Vasyl Purchel | 33m |
Runs and manages 23 clusters across five data centers with more than 1,200 servers and almost 100 PB raw using Ceph RBD/S3, containerized services and safadm… | ||
| Client Driven Geo-Redundancy for Ceph Object Storage | Yuval Lifshitz | 18m |
A Ceph RGW extension adds a new object_synced_create bucket-notification event that is emitted from the destination zone (configured on Zone B) when… | ||
| Optimizing Ceph IO for High Throughput Particle Physics Workflows | Tom Byrne | 43m |
65 PB Ceph cluster backing LHC analysis sustains ~20 GB/s but small vector reads against EC pools cause severe lock contention and 5+ second latencies for… | ||
| From Classical to the Future | Matan Breizman | 34m |
Project builds a new Crimson OSD rewrite using Seastar futures/promises to cut kernel context switches into cheap user-space task switches, targeting per-core… | ||
| Ceph for Public Cloud Workloads | Philip Williams | 21m |
Claim: colocating a Ceph cluster next to public-cloud regions cuts multi‑year storage bills dramatically while keeping fast internet access and control… | ||
| Building and Scaling Data Pipelines for Data Engineering With Ceph Object Storage | Kyle Bader | 35m |
Ceph RADOS Gateway sends S3 bucket notifications to Kafka, AMQ, or REST and triggers Knative-triggered serverless functions that scale to zero for realtime… | ||
| Maximising Data Storage Efficiency with Ceph as a Caching Layer for Tape Archiving | Thomas Bennett, Martin Slabber | 41m |
A working prototype uses ceph librados+rgw as an S3 interface, an NVMe cache and a separate Blue data manager that directs a blob-store to move objects onto… | ||
| Ceph Security BoF | JC Lopez | 36m |
Ceph and Rook now integrate IBM’s PSIRT and toolchain and claim stricter release security while preserving upstream compatibility; deployable encryption… | ||
| Implementing a Resilient Ceph RGW Object Storage with Erasure Coded Data Pool as the Foundation... | Siavash Sardari, Vahideh Alinouri, Somayeh Hajahmadi | 26m |
Digikala builds a Ceph RGW object store on reclaimed hardware with an erasure coded data pool instead of full replication, choosing a 2+4+2 erasure profile… | ||
| Bug of the Year: A Memory Leak Reboot Could Not Help with | Radoslaw Zarzynski | 42m |
A Ceph OSD memory leak stems from corrupt PG dupes entries that persist on disk so reboot does not help and an OSD must be redeployed to recover; the root… | ||
| Best UADK Acceleration Practice on Ceph Storage | Dai Zhiwei | 13m |
UADK integration offloads Ceph encryption and compression into hardware using IOMMU + shared virtual address to avoid memory copies and enforce device-boundary… | ||
| Optimizing CephFS with Combining MDS QoS Scheduling and Static-Dynamic Subtree Partitioning | Yongseok Oh, Jinmyeong Lee | 39m |
Combines an external worker static partitioner (writes client traces to Elasticsearch, pins subtrees rarely) with a balance rank mask for the dynamic balancer… | ||
| It’s Alive! Image Live Migration, Live Import and Instant Restore with Ceph RBD | Danny Harnik | 42m |
Image live migration moves a volume so the target immediately handles writes while reads fetch missing blocks from the source, using libvirt RBD only and a… | ||
| Improving Business Continuity for an Existing Large Scale Ceph Infrastructure | Enrico Bocchi, Abhishek Lekshmanan | 45m |
CERN runs 16 production clusters totaling ~65 PB and expects >100 PB; block storage scales from one to five RBD clusters exposed via OpenStack with storage… | ||
| Scrubs Scheduling in Reef and Beyond | Ronen Friedman | 51m |
Quincy introduces a new scrub scheduler that lands in Reef: each PG now stores separate shallow/deep targets instead of flag blobs, the queue sorts by a single… | ||
| Ceph QoS Refinements for Background Operations Using MClock | Sridhar Seshasayee, Aishwarya Mathuria | 44m |
Implements MClock QoS in Ceph OSDs, replaces static WPQ costs with a dynamic cost model (device random IOPS, sequential bandwidth, plus average object size per… | ||
| New Read Balancer in Ceph | Laura Flores | 37m |
A new read balancer for Reef adds a rebalance score shown by ceph osd pool ls detail and an offline CLI path using the OSD map tool that outputs pg primary… | ||
| How to Make a Positive Impact from Both an Environmental and Financial Perspective | Wout van Heeswijk, Wido den Hollander | 42m |
A concrete claim: a typical Dell-style rack server carries ~8,617 kg CO2e (100-year LCA) and running newer, slightly more efficient machines saves only ~48 kg… | ||
| Stretch Clusters in Ceph: Algorithms, Use Cases, and Improvements | Greg Farnum, Kamoltat (Junior) Sirivadhna | 32m |
Implements a two‑site stretch cluster mode for Ceph that enforces two copies per site, restricts OSDs to local monitors, and adds a tiebreaker monitor… | ||
| Critical Troubleshooting Tools for Rook Clusters | Subham Kumar Rai, Parth Arora | 35m |
A kubectl krew plugin called rook-ceph installs via krew and exposes one-line commands: rook status (prints Ceph and CR status and capacity), purge-osd… | ||
| A Welcoming Ceph Experience Improvements for First-Time Users or: How to Make Beginners Love Ceph | Jonas Sterr | 28m |
Deploys and ops notes: roughly 75 small clusters totalling ~225 nodes (≈5,400 cores, 58 TB RAM, 1.8 PB NVMe, ~900 OSDs) and consistent three-node… | ||
| Monitoring and Centralized Logging in Ceph | Avan Thakkar | 31m |
Shows a concrete stack: Prometheus + Alertmanager + Grafana for metrics and Loki + Promtail for centralized logs, with dashboards embedded in the Ceph… | ||
| Over a Billion Requests Served per Day: Ensuring Everyone Is Happy with Our Ceph Clusters’... | Nathan Hoad, Alex Wojno | 34m |
Bloomberg runs over a billion S3 requests per day across four Ceph sites (~65 PB, ~4,500 OSDs, ~50k buckets) and builds a software-defined QoS on the RGW that… | ||
| Ceph in Scientific Computing and Large Clusters BoF | Kevin Hrpcek | 39m |
BoF reports a production Ceph front end that serves a 30 PB cluster with ~3,000 OSDs and exposes only S3, using an in‑house frontend and backend to mint 4‑hour… | ||
| BlueStore V2 - an Evolutionary Step Forward | Adam Kupczyk, Mark Nelson | 46m |
BlueStore V2 describes concrete fixes and new components: a lazy unpinning approach that stabilizes cache reference counting after nine PR attempts, a bootable… | ||
| A 10-Year Retrospective Operating Ceph for Particle Physics | Dan van der Ster | 38m |
Started with a 300 TB proof of concept in 2013 and grows to ~100 PB raw across 17 Ceph clusters, with an early jump to 3 PB in production; experiments include… | ||
| Ceph CSI Driver: Bridging Containers and Ceph | Rakshith R, Yati Padia | 40m |
Ceph CSI implements gRPC CSI servers in Go to let Kubernetes (and other orchestrators) provision RBD block volumes, CephFS subvolumes and NFS exports (RBD… | ||
| Rook: Enabling Read Affinity for RBD Workloads | Rakshith R | 19m |
Rook implements read affinity for RBD by passing node topology labels as CRUSH locations through the CSI driver to krbd mount options so clients read from the… | ||
| Ceph RBD Integration with DPUs | Varada Raja Kumar Kari | 28m |
Prototype runs a mini librbd on DPU that exposes NVMe namespaces to hosts and converts IO to RADOS calls, using AMD/Pensando ARM cores plus MPUs for… | ||
| Making CephFS (Much More) Great! | Patrick Donnelly, Greg Farnum | 37m |
Session reports that multi‑MDS recovery remains fragile but offers concrete fixes: an option to bring whole FS down for safe upgrades, a proposed CLI wrapper… | ||
| Secure Token Service in Ceph | Pritha Srivastava | 22m |
Ceph RGW implements AWS‑style STS: it issues encrypted temporary credentials (access key, secret, session token) and supports assume role, assume role with web… | ||
| Optimizing Ceph Object Storage for Analytics and AL/ML Workloads - Overview and Future Plans | Matt Benjamin | 36m |
Ceph demonstrates concrete scale and performance: tests hit 10 billion objects on six nodes and a single-bucket run holds 250 million objects, while a 10-node… | ||
| Enabling Near-Data Processing in Ceph | Rixin Luo | 20m |
Adds an aggregate module that groups same-pool small objects into a single stripe-aligned aggregate object so no original object is split before erasure… | ||
| Ceph Performance Tuning: From Bluestore to RBD | Mark Nelson | 38m |
Uses CBT Upstream and fio/hsbench for repeatable tests on a 60‑OSD NVMe cluster and shows concrete wins: reads hit 4.5 million IOPS while burning ~8 cores per… | ||
| An Initial Look at Deep Learning IO Performance | Mark Nelson | 37m |
Benchmarks use ImageNet (1.3M images, ~160 GB) and TFRecords (about 1,000 files ~1.16 GB each) on Lambda Labs instances with 8 V100 GPUs and stock setups… | ||
| Ceph Crimson Verification on AArch64 | Rixin Luo | 13m |
Built ARM Crimson images and deployed a three-node cluster with cephadm + fio to benchmark c-store; discovered a memory-leak pattern in buffer lists that can… | ||
| nothing matches | ||