On Thu, 2020-04-23 at 09:08 +0200, Janne Johansson wrote: > Den tors 23 apr. 2020 kl 08:49 skrev Darren Soothill < > darren.soothill@xxxxxxxx>: > > > If you want the lowest cost per TB then you will be going with > > larger nodes in your cluster but it does mean you minimum cluster > > size is going to be many PB’s in size. > > Now the question is what is the tax that a particular chassis > > vendor is charging you. I know from the configs we do on a regular > > basis that a 60 drive chassis will give you the lowest cost per TB. > > BUT it has implications. Your cluster size needs to be up in the > > order of 10PB minimum. 60 x 18TB gives you around 1PB per node. Oh > > did you notice here we are going for the bigger disk drives. Why > > because the more data you can spread your fixed costs across the > > lower the overall cost per GB. > > > > I don't know all models, but the computers I've looked at with 60 > drive slots will have a small and "crappy" motherboard, with few > options, not many buses/slots/network ports and low amounts of cores, > DIMM sockets and so on, counting on you to make almost a passive > storage node on it. I have a hard time thinking the 60*18TB OSD > recovery requirements in cpu and ram would be covered in any way by > the kinds of 60-slot boxes I've seen. Not that I focus on that area, > but it seems like a common tradeoff, Heavy Duty(tm) motherboards or > tons of drives. I would imagine that this describes the use of separate SAS-attached (or whatever) JBOD boxes rather than everything in a single chassis. My clusters use 1U servers with decent CPU/memory and SAS adapter cards hooking up larger JBODs to actually house the disks (for the spinning rust OSDs, at least).
Attachment:
signature.asc
Description: This is a digitally signed message part
_______________________________________________ ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an email to ceph-users-leave@xxxxxxx
