The Must-Attend

Open Infrastructure Event

The world runs on open infrastructure. At the OpenStack Summit, you’ll learn about the mix of open technologies building the modern infrastructure stack, including OpenStack, Kubernetes, Docker, Ansible, Ceph, OVS, OpenContrail, OPNFV, and more. Whether you are pursuing a private, public or multi-cloud approach, the OpenStack Summit is the place to network, skill up and plan your cloud strategy.

Boston, MA
May 8-11, 2017

The Must-Attend

Open Infrastructure Event

Join the movement behind the most widely deployed open source cloud platform. Whether you are pursuing a private, hybrid or multi-cloud approach, the OpenStack Summit is the place to network, skill up and plan your cloud strategy.

Boston, MA
May 8-11, 2017

Event Details


Unlocking the Performance Secrets of Ceph Object Storage

Object storage is clearly the simplest, most flexible, and most scalable way to store tons of unstructured content. In the last few months we were on a journey to discover "The Unknowns" from the world of Object storage. We have conducted an extensive testing with more than 1500 unique benchmarking tests across a wide range of hardware/software configuration. We have a ton of interesting things to share , so that you can design the best in class Ceph object storage solution for your workloads.

Join the experts to unlock the performance secrets, and take away answers to some of the most fundamental questions :

  • How to achieve max small/large, object (throughput/bandwidth) on fixed cluster size?
  • What happens at workload saturation point and how does the machinery behaves ?
  • What effect does a caching layer have on performance ? ( Hint : Never seen before results )
  • How much effect does reducing or eliminating index operation overhead have on small object throughput ?

What can I expect to learn?

Attendees will learn the skills of benchmarking, tuning and best practices for Ceph Object Storage performance tuning. They will also take away answers to some of the most fundamental questions :-

  • How to achieve max small object operations on fixed cluster size?
  • How to achieve max large object throughput on fixed cluster size?
  • What happens at workload saturation point and how does the machinery behaves ?
  • What effect does data protection schemes have on price/performance?
  • What effect does total cluster object count have on workload performance?
  • What effect does a caching layer have on performance ? ( Hint : Never seen before results )
  • How much does performance improve with large numbers of small objects when using
  • Bucket index sharding
  • Indexless Buckets
  • Bucket Index on Flash Media vs. Spinner
Wednesday, May 10, 1:50pm-2:30pm
Will be recorded
Level: Intermediate
Red Hat
Karan Singh is working as senior storage architect at Red Hat based in Finland. At Red Hat his responsibilities include designing, implementing and testing various reference architectures, performance and sizing guides based on Ceph. He is deeply involved in Ceph performance evaluation across wide range workloads and industry standard hardware from different providers. Karan has been working... FULL PROFILE
Red Hat, Inc
Kyle Bader is a Senior Solution Architect working in the Storage Solutions Team at Red Hat, lending his design and operational skills with Ceph to help develop tested solutions that ensure repeatable success when deploying distributed, fault-tolerent, multi-petabyte storage systems. Prior to Red Hat, Kyle had architectural roles at both Inktank and DreamHost. Kyle was part of the team that... FULL PROFILE
Red Hat Inc
Brent Compton is Director Storage Solution Architectures at Red Hat, leading the team responsible for building Ceph and Gluster storage reference architectures with Red Hat Storage partners.  Before Red Hat, Brent was responsible for emerging non-volatile memory software technologies at Fusion-io.  Previous enterprise software leadership roles include VP Product Management at... FULL PROFILE