In this section we'll explore strategies for placing your Ceph servers within a data center. Consider in advance each of these factors to prevent potentially expensive (and embarrassing) rude awakenings after you deploy your production clusters.
Literature and the media have been buzzing for a few years about Converged Infrastructure and as if that weren't catchy enough, Hyperconverged Infrastructure. Perhaps, next year we'll see Ultraconverged Infrastructure that also makes julienne fries. The distinction between the two is mostly marketing. Traditional cloud and server farm designs have typically featured servers / chassis dedicated to each role: Compute, Storage, Networking. Convergence however is all the rage today: provisioning all services onto the same servers. The touted advantages of this approach are savings in management costs and maintenance (read: headcount), in DC footprint recurring operational costs, and in hardware up front capital...