Skip to main content

Your submission was sent successfully! Close

Thank you for signing up for our newsletter!
In these regular emails you will find the latest updates from Canonical and upcoming events where you can meet our team.Close

Thank you for contacting us. A member of our team will be in touch shortly. Close

  1. Blog
  2. Article

Philip Williams
on 22 November 2021

Dell EMC PowerEdge and Canonical Charmed Ceph, a proven solution


Here at Canonical, we have lots of industry partnerships where we work jointly, hand-in-hand, to produce the best possible outcomes for the open source community. From getting early access to next generation hardware to ensure Ubuntu is fully compatible when it’s released, to creating solution orientated reference architectures for products built on top of Ubuntu like Charmed Ceph, Canonical is committed to engineering the best possible computing experience.

Recently, our product management and hardware alliances teams came together with Dell Technologies to collaboratively define, test, and validate a Dell EMC PowerEdge based Charmed Ceph reference architecture.

Reference architecture

The goal of this exercise was to produce a guide to building a capacity orientated Ceph cluster that could be used for block (RBD), file (CephFS) or object (Swift or S3) workloads, and demonstrate the performance that can be achieved with similar hardware.

We took relatively standard components (four Dell EMC R740xd2 servers with Intel Xeon processors and NICs, a few SSDs, and lots of high capacity NL-SAS disks) and connected them all together with 25GbE networking.

The R740xd2 provides an ideal building block for Ceph clusters due to its highly configurable nature, which allows users to make performance, capacity, and price adjustments as needed. For example, to create a higher performance cluster, the CPUs could be swapped for another model that has more cores and cache, and the disks could be changed to NVMe and/or SSD if required.

Learn more

During this exercise, we tested the performance of the cluster with various different workloads, such as small block and large block, with and without bcache.

We also demonstrated the scalability of Ceph, by adding an extra storage node and re-running the performance tests to show the improvement in cluster performance. We were able to achieve over 75,000 random read IOPs, and over 6 GBps sequential read from a 4 node capacity orientated cluster, as well as demonstrating how our unique OSD deployment approach using bcache can provide up to 2.5x improvement in performance for small block workloads.

All of the test results and detailed hardware architecture information can be found in the whitepaper on Dell Technologies InfoHub, here. We also discussed our findings in this webinar which can be watched back on-demand.

Related posts


Philip Williams
16 August 2024

Managed storage with Ceph

Ceph Article

Treat your open source storage infrastructure as a service What if storage was like coffee: menu driven and truly service oriented? Everyone knows how quick and easy it is to order a cappuccino or cortado and have a friendly barista bring it to you in just minutes. Now imagine this is a user who needs ...


Philip Williams
25 July 2024

How do you select the best enterprise data storage solution for your business?

Ceph Article

The choices you make around IT infrastructure have great impact for both business cost and performance, across areas as diverse as operations, finance, data analysis and marketing. Given the importance of data across all of these areas and frankly, across your business as a whole, making the right decision when choosing a new storage syst ...


Philip Williams
16 July 2024

The guide to cloud storage security for public sector

Ceph Article

Cloud storage solutions can provide public sector organisations with a high degree of flexibility when it comes to their storage needs, either public cloud based, or in their own private clouds. In our previous blog post we looked at the economic differences between these two approaches. In this blog we will explore some of the ...