Cephalocon 2019

Benchmarking Ceph is conceptually easy: allocate compute resources, build and deploy Ceph, run benchmarks, gather and analyze results. In practice, and especially for people getting started with Ceph development and benchmarking, implementing this type of experimentation pipeline is a time-consuming effort. In this talk, Ivo Jimenez will present a community-based effort for creating and maintaining Ceph experimentation pipelines. These are used at UC Santa Cruz in research projects and classes to build, deploy and test Ceph in R&D scenarios. These pipelines follow DevOps principles (e.g. infrastructure-as-code) to allocate bare-metal resources, deploy Ceph using ceph-ansible, and run tests using CBT. Thanks to this effort, we are able to smooth the learning curve for students and others interested in starting to "get their hands dirty" with Ceph.