Cephalocon Barcelona 2019
2019-05-19..2019-05-20
Barcelona, Spain
| title | author | length |
|---|---|---|
| Keynote: State of the Cephalopod | Sage Weil | 36m |
Ceph updates to Nautilus include an integrated GUI dashboard and an orchestrator API that automates deployment via Rook, Ansible or SSH; PG autoscaling that… | ||
| Keynote: What's Planned for Ceph Octopus | Sage Weil | 18m |
Octopus focuses on concrete features: a new orchestrator API that exposes add/replace OSD, provision monitors, and drive deployments to both CLI and dashboard… | ||
| Keynote: Town Hall | 35m | |
Panel summarizes concrete next steps and contribution paths for Ceph: prioritize the backporting effort to learn the codebase, use the new edit on GitHub links… | ||
| Failing Better - When Not To Ceph and Lessons Learned | Lars Marowsky-Brée | 41m |
Ceph excels at hyperscale but struggles in small or single-threaded deployments: a four-node POC often underperforms and offers little economy of scale, 20 TB… | ||
| Ceph Orchestrator: Bridging the Gap Between Ceph and Deployment | Sebastian Wagner | 41m |
Ceph Manager exposes a single API that maps actions across Ceph-Ansible, DeepSea, Rook so cluster state and service operations look the same to tooling. The… | ||
| Ceph Manager Dashboard - The New Way To Manage Ceph & Gateway Management in Ceph Dashboard | Kai Wagner, Ricardo Marques, Ricardo Dias | 37m |
Dashboard now exposes live cluster management: user roles and SSL, RADOS RGW Admin Ops API for S3 users/buckets, pool and OSD controls (set flags, mark out… | ||
| A Glimpse of the New Ceph Messenger Built on Seastar | Yingxin Cheng, Vivian Zhu | 37m |
A new Ceph messenger (Crimson) implements Seastar’s shared‑nothing, future/continuation model, integrates POSTMESSAGE v1 and v2, and is merged with Crimson… | ||
| Rebuilding a Faster OSD with Future | Kefu Chai, Radoslaw Zarzynski | 27m |
Crimson rewrites the Ceph OSD IO path with Seastar futures to eliminate locks and pin one OSD thread per core, targeting NVMe-class devices (modern SSDs ~0.5M… | ||
| CephFS as a Scalable Filer | Rafael Fernández López, Brett Milford | 40m |
Monash runs CephFS as a production filer on ~100 servers, ~2000 OSDs with two data pools (216 and 342 OSDs), a carved SSD metadata pool, five front-end VMs and… | ||
| Configuring Small Ceph Clusters for Optimal Performance | Josh Salomon | 42m |
Three‑node Ceph cluster (40 Gb NICs) tests show out‑of‑the‑box defaults underperform and small clusters need tuned PG counts, balancing and sizing. Using… | ||
| Exploring the Performance Limits of CephFS in Nautilus | Manoj Pillai | 31m |
Benchmarking Nautilus CephFS with FIO, smallfile and a harness called dimmer shows clear limits: NVMe drives deliver ~400k 4K random IOPS and ~2–5 GB/s… | ||
| Fresh from the Labs: Containerized OpenStack and Ceph Performance | Karan Singh, Rachana Patel | 34m |
Fresh from the Labs: Containerized OpenStack and Ceph Performance Head-to-Head (HCI vs. De-Coupled) - Karan Singh & Rachana Patel, Red Hat
Containerized… | ||
| Optimize librbd for Lower CPU Cost and Higher Scalability | Li Wang | 26m |
Replaces single-buffer writes with rbd_aio_writev and implements zero copy in the librbd messenger, removing two memcpy hotspots so 1 MiB sequential writes… | ||
| Making Ceph Fast in the Face of Failure | Neha Ojha | 35m |
Ceph adds concrete controls and algorithms to keep client I/O responsive during recovery: default OSD recovery sleep = 0.1s (HDD), 0s (SSD), 0.025s (hybrid)… | ||
| Testing Ceph for the Cloud, in the Cloud | Adam Wolfe Gordon | 34m |
DigitalOcean runs Ceph for block and S3-compatible object storage across ~21 clusters in 9 datacenters (~40 PB) and automates integration testing by… | ||
| Juggling Petabytes: Managing Ceph at Scale with Ceph-ansible | Matthew Vernon | 40m |
Operates a single Ceph cluster built from Supermicro nodes with 18 PB raw capacity and about 40% used, deployed and managed with ceph-ansible plus local… | ||
| Day 2 Operations : Make Friends with Your Ceph Cluster | Adrien Gillard | 34m |
Assumes a healthy cluster and then gives concrete operational rules: aggregate logs with ELK stack and tune log levels; collect metrics with Prometheus +… | ||
| Unlimited Fileserver with Samba CTDB and CephFS | Robert Sander | 42m |
A practical guide to building a scalable, highly available file service by combining Samba clustered with CTDB and CephFS as the backend. It covers configuring… | ||
| RADOS Object Class Development in C++ Lua | Noah Watkins | 30m |
Shows how Ceph object classes let compiled C++ modules or Lua scripts run inside OSDs to perform work like MD5 checksums or ImageMagick thumbnailing so only… | ||
| Per OSD Recovery Bandwidth Control Based on dmClock | Xie Xingguo, Yan Jun | 20m |
Prototype implements dmClock to enforce per-OSD bandwidth limits for recovery, exposing three fixed levels (low/medium/high) plus a self-adaptive mode that… | ||
| Unified Cluster Federation: A Hyperscaled Architecture of... | Varada Raja Kumar Kari, Romit Misra | 46m |
Unified Cluster Federation: A Hyperscaled Architecture of Heterogenous Clusters - Varada Kari & Romit Misra, Flipkart
Flipkart is a leading Indian e-commerce… | ||
| A Hitchhiker's Guide to Ceph RGW PubSub & Cats vs. Dogs - a Hybrid Cloud Storage Story | Huamin Chen, Yehuda Sadeh-Weinraub, Yuval Lifshitz | 34m |
A Ceph RGW sync-module implements pub/sub that stores events in Ceph and forwards them via HTTP and AMQP to existing brokers; configuration uses topics and… | ||
| Affordable NVMe Performance on Ceph & Ceph on NVMe - True... | Wido den Hollander, Piotr Dalek | 45m |
Ceph on NVMe works if planned: expect ~0.7–0.8 ms 4K write latency with replica 3 and tuned CPUs, but vendor whitepapers often use replica‑1/2, top‑end CPUs… | ||
| Optimizing Ceph Object Storage for Production in Multisite Clouds | Michael Hackett, Vikhyat Umrao | 44m |
Sizing and tuning advice for Ceph RGW clusters with concrete knobs: choose media by workload (NVMe/SSD for index/DB, HDD for cold), size BlueStore DB ~4% of… | ||
| Healthier Ceph Clusters with Ceph-medic | Alfredo Deza | 05m |
Ceph-medic automates a short checklist of roughly 20 quick health checks across deployments (supports bare metal, containers, OpenShift/Kubernetes) by… | ||
| Using Devops Practices for Operating CEPH | Anders Bruvik | 07m |
Argues that DevOps practices apply directly to operating a CEPH cluster and shows what that looks like: store configuration in version control, deploy with CI… | ||
| Ceph BenchOps: Following DevOps Practices for Benchmarking Ceph | Ivo Jimenez | 05m |
Classes have three-month quarters so setup time matters. The project provides reproducible, end-to-end pipelines that GitHub Actions workflows containerize… | ||
| Importance of CRUSH - A Life Lesson | Gaurav Sitlani | 08m |
Importance of CRUSH - A Life Lesson - Gaurav Sitlani, Red Hat
This talk is all about DO's and Dont's when someone tries to play with CRUSH , explaining from… | ||
| I Need More Space, It's Not You It's BlueStore | Mohamad Gebai | 06m |
Cluster-report mismatch: ceph df shows 76 GB max and 17 GB stored but 92% used because one OSD is 94% full; solution is rebalance with upmap balancer… | ||
| RWX Storage for Container Orchestrators with CephFS and Manila | Tom Barron | 41m |
Goal is to deliver production RWX volumes for Kubernetes (and other orchestrators) by wiring CephFS into OpenStack Manila and exposing it via a CSI driver… | ||
| Rook Deployed Scalable NFS Clusters Exporting CephFS | Patrick Donnelly, Jeff Layton | 34m |
Rook and Kubernetes deploy NFS-Ganesha pods that export CephFS; Nautilus adds a RADOS-backed coordination store so RADOS epoch coordination keeps atomic epoch… | ||
| Ceph Operations at CERN: Where Do We Go From Here? | Dan van der Ster, Teo Mouratidis | 31m |
CERN runs roughly 20 PB of Ceph and reports concrete results: a 3 PB live migration with zero downtime, filestore→bluestore moves that cut CPU load and doubled… | ||
| Object Bucket Provisioning in Rook-Ceph | Jonathan Cope, Jeff Vance | 28m |
A Kubernetes operator library implements a generalized S3 bucket API for Rook-Ceph and other object stores, exposing CRDs that include ObjectBucketClaim (OBC)… | ||
| Rook - Running Ceph Using Kubernetes | Alexander Trost, Kim-Norman Sahm | 41m |
Rook v0.9 automates provisioning of a full Ceph cluster inside Kubernetes: the operator creates 3 monitors, manager, and OSD prepare jobs, configures block… | ||
| Keynote: Pushing the Limits of Ceph Performance through Software & Hardware Innova... | Tushar Gohad | 12m |
Intel describes concrete upstream work: SPDK/DPDK and PMDK backends for SSDs are under development, BlueStore and Rocks improvements land, and Seastar-based… | ||
| Keynote: Ceph as Part of the Data Infrastructure for Zoned Storage | Jorge Campello De Souza | 10m |
Ceph integration targets Zoned Block Devices such as SMR HDDs and emerging NVMe ZNS namespaces by exposing zone semantics to the host to cut write… | ||
| Keynotes: Supermicro® SuperStorage Systems Based on New Intel® Xeon® Scalable Pro... | David Ramirez | 09m |
Supermicro SuperStorage systems pair Cascade Lake Xeon processors with NVMe and Intel Optane DC Persistent Memory to cut cost and boost speed — claims include… | ||
| Keynote: Ceph Journey - A Perspective | Dr. Gerald Pfeifer | 17m |
SUSE describes hard wins: a University of Maine Ceph cluster stores 2 million files in one directory and halves the workload runtime; a large tech team… | ||
| Keeping up a Competitive Ceph/RadosGW S3 API | Javier Muñoz | 26m |
Practical account of contributing features and bugfixes to RadosGW S3 across releases, showing what lands upstream and what actually changes product behavior… | ||
| Messenger V2: The New Ceph Wire Protocol | Ricardo Dias | 37m |
Ceph ships a redesigned wire protocol, Messenger V2, implemented for Nautilus that adds extensibility, pluggable authentication and optional on‑the‑wire… | ||
| Geographical Redundancy with rbd-mirror: Best... | Florian Haas | 38m |
rbd-mirror (introduced in Jewel) replicates RBD by consuming a client-side rbd journal and replaying it to a remote image; it supports one-way and two-way… | ||
| Object WORM Feature in Ceph Rados Gateway | Zhang Shaowen | 15m |
Ceph radosgw gains S3-compatible WORM via an Object Lock implementation that China Mobile contributes: a prototype adds six new APIs to set bucket defaults and… | ||
| Monitoring Ceph with Prometheus | Jan Fajerski | 33m |
Shows how to run Prometheus against Ceph using the Ceph manager exporter, node exporter and file_sd for dynamic targets, and how to correlate labels to join… | ||
| echo "Subject: Update on librmb" | sendmail -v SDS@ceph.com | Danny Al-Gaaf | 40m |
Deutsche Telekom describes librmb, a library that stores email directly in Ceph RADOS objects on a petabyte-scale proof-of-concept cluster to enable parallel… | ||
| 200 Clusters vs 1 Admin | Bartosz Rabiega | 36m |
OVH runs ~210 Ceph clusters totalling ~50 PB on ~2,000 servers and 25,000 HDDs, and keeps one admin effective over 200 clusters by building a centralized… | ||
| What are “caps”? (And Why Won’t my Client Drop Them?) | Gregory Farnum | 34m |
CephFS uses metadata caps (not CephX keys) that the MDS issues—examples: pin cap, auth cap, link cap, file cap—each with bits like shared/exclusive… | ||
| RGW S3: Feature Progress, Limitations & Testing | Robin H Johnson, Ali Maredia | 33m |
Status update on RGW S3 compatibility, costs, and testing: RGW implements many S3 features but gaps exist (append-object, PubSub divergence, partial lifecycle… | ||
| Running Backups with Ceph-to-Ceph | Michel Raabe | 29m |
Compares concrete options: RBD mirror for asynchronous image replication (requires dual-cluster connectivity, default 30‑second sync window so deletions… | ||
| Avoiding the Curious Case of Leaky Cauldrons | Abhishek Lekshmanan | 24m |
Avoiding the Curious Case of Leaky Cauldrons - Abhishek Lekshmanan, SUSE Linux Gmbh
A large number of leaks in S3 are caused by users accidentally setting… | ||
| Rapid Processing of NASA Satellite Data Stored with Ceph | Kevin Hrpcek, Steve Dutcher | 36m |
A six-petabyte Ceph cluster using librados, EC pools (6+2 erasure coding) and CephFS stores millions of VIIRS granules and ingests ~600 GB/day while producing… | ||
| Releasing Krakens & Other Cephalopods Into the Wild | Abhishek Lekshmanan | 23m |
Releasing Krakens & Other Cephalopods Into the Wild - Abhishek Lekshmanan, SUSE
Every major release of Ceph comes with an approximate 18 month lifecycle with… | ||
| Testing Ceph: Status, Development, & Opportunities | Gregory Farnum | 36m |
Ceph testing runs on the teuthology framework driving a 250+ server lab and thousands of machine‑hours of daily jobs; it combines combinatorial suite… | ||
| Releasing Ceph - Deep Dive Into Build Infrastructure | Alfredo Deza, Ken Dreyer | 35m |
A release pipeline migrates from monolithic per-distro servers and exposed GPG keys to automated ephemeral build nodes and a restricted signing station… | ||
| Getting Started as a Rook-Ceph Developer | Blaine Gardner | 34m |
Step-by-step developer flow for Rook Ceph: build the Ceph-only container, push images to nodes via docker save / scp / docker load, install manifests, run… | ||
| Hands On with Rook: Ceph & Kubernetes | Maxime Guyot, John Studarus | 01h 21m |
Hands-on workshop deploys Rook on Kubernetes over bare metal NVMe SSDs with 80 lab environments and two physical nodes, starting with a single-node Ceph… | ||
| Highly Available Git on CephFS with Rook, Kubernetes, and OpenStack | James E. Blair | 42m |
OpenDev replaces an eight-VM, shared-nothing git farm with a reproducible, shared-everything stack: Kubernetes on OpenStack bootstrapped by… | ||
| Keynote: Supporting Swiss Academia with Ceph & OpenStack | Jens-Christian Fischer | 23m |
SWITCH runs an OpenStack IaaS backed by multiple Ceph clusters in production since 2014, scaling from an eight‑node experiment to ~400 servers and 10 petabytes… | ||
| Keynote: A Wonderful Journey On Ceph | Luo Kexue | 08m |
ZTE describes building CloveStorage, a Ceph-compatible distributed storage product that integrates a smart cache, cache gateway and an automated… | ||
| Keynote: The System that Matters | Tim Massey, Phil Straw | 16m |
SoftIron announces silicon acceleration for Ceph erasure coding, moving from a naïve offload that showed no gain to a matched hardware‑OS‑driver design that… | ||
| CRUSH-ing the OSD Variance Problem | Tom Byrne | 44m |
Manages a Ceph cluster that grows from 1,500 to about 5,000 OSDs (≈40 PB raw) while maintaining >50% average utilization, and faces wide OSD-utilization… | ||
| MeerKAT Astronomy Data Store Deployment and Operations | Martin Slabber | 42m |
MeerKAT runs two Ceph clusters — a Cape Town archive with 2640 hard drives and 55 NVMe devices and an on-site Karoo cluster with 240 hard drives and 480 SSDs —… | ||
| Delivering Scalable Data Lake Analytics Solutions on Ceph with High Capacity Flash | Anjaneya "Reddy" Chagam | 49m |
Delivering Scalable Data Lake Analytics Solutions on Ceph with High Capacity Flash - Anjaneya Chagam, Intel Corporation
Cloud Service Providers (CSPs)… | ||
| Upgrade and Scale-Out an In-production Ceph Cluster on Mixed Arm Micro-server Platform | Aaron Joue | 40m |
Case study of an in-production Ceph+OpenStack cluster: starts with 21 OSDs serving ~140 VMs, cluster peaks ~7,000 IOPS with per-disk averages ~100 IOPS so… | ||
| Ceph Practice And Usage In China Mobile | Zhang Shaowen | 19m |
Ceph is in production since 2016 with hundreds of petabytes across block and object; block clusters exceed ~500 PB and single clusters run ~2,600 OSDs… | ||
| Brazilian Government Case | Brenno Martinez | 05m |
Brazilian digital driver licenses store 12 objects per license, including photo signature fingerprints and a QR code, totaling 600 million objects in Ceph. The… | ||
| Practices of Using NFS Over RGW | Enming Zhang | 06m |
Tests write two 1 GB files over NFS to RGW and observe only 1 GB reclaimed while 2 GB remains until GC ends, traced to duplicate GC tags sent repeatedly so… | ||
| Practices of Ceph Object Storage in Public Cloud Services | Yu Liyang | 06m |
Three-city multi-site Ceph deployment for China Mobile runs a public object-cloud with regional metadata sync and over 15PB of raw capacity, using Ceph RGW/S3… | ||
| Learn Ceph — For Fun, For Real, For Free! | Florian Haas | 06m |
City Cloud Academy offers a self-paced Ceph course (CC213) with fully interactive cloud labs that let learners build a Ceph cluster with Ceph Ansible, practice… | ||
| Configuring Ceph Deployments with an Easy to Use Calculator | Karl Vietmeier | 06m |
Empirical benchmarks on an all‑flash Ceph cluster (Broadwell servers, Intel NVMe 4500/Optane hardware, Bluestore, 3x replication, IO tests via CBT and Grafana… | ||
| Rapid Design and Effective Operating of a General Purpose Object Storage at RWTH | Jonas Jansen | 08m |
Collaborative Ceph deployment with four partners builds a geo-redundant S3 system across 4 geo-distributed data centers over dark fiber, adding NAS and… | ||
| Ceph and NVMe/TCP | Orit Wasserman | 04m |
NVMe/TCP is in Linux kernel 5.0 and delivers NVMe-over-fabric without changing network gear; integration with Ceph moves metadata off local disks so clusters… | ||
| Nobody Knows What PGs are Good For, Only I Do | Danil Kipnis | 06m |
Shows that PGs let a distributed block store trade recovery speed for safety using concrete numbers and setups: a tiny cluster of three machines with two disks… | ||
| nothing matches | ||