Amsterdam was the perfect setting for two of the world's biggest events related to communities Ceph and Kubernetes: Cephalocon and KubeCon.
This combination allowed me, and about 11 thousand people from all over the world, to expand their network of knowledge and skills related to topics ranging from storage (Ceph) to DevOps and Cloud Native (Kubernetes and everything that surrounds it). It all started on Sunday, April 16, the starting date of the Cephalocon and ended on Friday, April 22. 6 days full of incredible experiences.
Even if it won't be easy, here are the Highlights of the first event that welcomed me to Amsterdam: Cephalocon!
Ceph pushes towards Cloud Native
The first news presented at Day0 of the main conference that stars the Ceph Software-Defined Storage (SDS) was the acquisition of Red Hat by IBM and the consequent Radical change in the team that coordinates the Ceph Community. From day one, dedicated to the contributors of the project of which CloudFire is a part, it has been clear that New road that IBM wants to undertake with the development: Ceph must revolutionize to meet the needs created by the massive adoption of Kubernetes!
Now considered a “commodity”, the most used open source cloud storage wants to partially change course in order to adapt to the new workloads that the Cloud Native world requires. How will this translate? The guidelines are related to the self-service and DevOps management of RBD volumes and to the performance and security side of the S3 part (RadosGateway).
The name of the new release was then announced: Squid (19.0.0 - currently under design), which will be the one after Reef (18.0.0 - currently NOT stable) which will bring significant performance improvements and will introduce support for NVMe-oF, which will be easily backported at least to the Reef release, now close to being promoted to Stable.
Performance and Stability
An important step forward has also been taken with the Next-Gen OSDs, and with the Crimson project (currently NOT production ready, but in the last phase of testing) Change the paradigm in metadata management, going to solve the problem that has always afflicted the OSD/BlueStore architecture, namely the physiological latencies introduced by the management of Metadata during writing. The Crimson project sees, among the various beta-testers, also bodies such as Bloomberg and CERN that have brought very interesting talks on the previews of Crimson's performance and stability with OSD NVMe.
Final Thoughts
Cephalocon is also an opportunity for end-users of confronting on topics such as optimization and debug of problems, especially with the various use cases. In this sense, events such as Cephalocon are essential to be able to get feedback from bodies such as CERN in Geneva or Bloomberg where best practices are literally written down.
An incredible opportunity for growth that comes after 4 years of stoppage and we are sure that, with the entry of IBM and with the new vision that has been provided, Cephalocon will once again be a central event in the Ceph community of which CloudFire is proudly part both as an End-User and as a Contributor! 🐙
