Designing Ceph clusters is hard. Throw OpenStack into the mix, and it gets even harder. There are many different ways in which a Ceph clusters can be configured. Same goes for selecting the right hardware. This session will give you a solid start into the world of Ceph-based storage (in the context of OpenStack).
You will learn all the lessons learned by us the hard way when operating our two Ceph clusters over the past several years including all necessary recommendations on picking the right hardware for Ceph, as well as some tips on optimising the deployed Ceph cluster for OpenStack.
During the session will concentrate on what to consider when creating your very own Ceph cluster, what are the common problems with Ceph to be aware of, what is benchmarking and profiling Ceph, what to know about fat node vs thin node, what’s the best SSD for journal (Spoiler alert: one does not exist, here’s why) and, the last but not least, what is the sweet spot between storage density, price, and performance.
The session will be a good blend of high level information combined with various useful to know technical details which we’ve learned the hard way, so that you don’t have to.
