Every Thursday the Ceph community convenes to discuss the ongoing performance work related to Ceph. The collaborative work by a number of different individuals and organizations is what has helped Ceph performance to come so far in such a short amount of time. Please join us if you are interested in all things Ceph performance.
When: Every Thu, 8am PST
Where: Blue Jeans teleconference
Recordings: All calls will be recorded for later viewing.
Mailing list: to receive announcements, watch the ceph-users list.
Intercall Toll Free Dial-In Number (US & Canada): (888)847-2201
List of other dial in numbers: Red Hat Conf Numbers
Conference code: 944 567 299
CBT is a testing harness written in python that can automate a variety of tasks related to testing the performance of Ceph clusters. CBT records system metrics with collectl, it can optionally collect more information using a number of tools including perf, blktrace, and valgrind. In addition to basic benchmarks, CBT can also do advanced testing that includes automated OSD outages, erasure coded pools, and cache tier configurations.
A number of use cases and reference architectures are available to the Ceph community to see what other experts are doing to setup and tune their clusters for particular workloads. Take a look and see if there is something that can give a head start to your own performance tuning efforts.