On Fri, Nov 4, 2011 at 1:12 AM, Amon Ott <a.ott@xxxxxxxxxxxx> wrote: > I agree that our needs are special. We want to be able to start with 2 and > extend as customer needs grow, or start with 20 and shut some of them down if > load gets low. I now understand that we should start with at least three > nodes. It's not so special a usage case, you're just thinking of it wrong. Adding a new daemon of each type for every physical box you add is the wrong way to expand. You want to add OSDs for every box in order to take advantage of the storage; you CAN add an MDS for each box, but you should really be scaling them according to how much metadata activity you need to support; and the monitors need to be handled carefully in order to keep a strict majority of them alive at all times (and aside from durability the system doesn't much care about how many there are). -Greg -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
