Hi Scott, Just some observations from here.
We run 8 nodes, 2U units with 12x OSD each (4x 500GB ssd, 8x 4TB platter) attached to 2x LSI 2308 cards. Each node uses an intel E5-2620 with 32G mem. Granted, we only have like 25 VM (some fairly io-hungry, both iops and throughput-wise though) on that cluster, but we hardly see any cpu-usage at all. We have ~6k PG and according to munin our avg. cpu time is ~9% (that is out of all cores, so 9% out of 1200% (6 cores, 6 HT)). Sadly I didn't record cpu-usage while stresstesting or breaking it. We're using cuttlefish and XFS. And again, this cluster is still pretty underused, so the cpu-usage does not reflect a more active system. Cheers, Martin On Mon, Oct 7, 2013 at 6:15 PM, Scott Devoid <dev...@anl.gov> wrote: > I brought this up within the context of the RAID discussion, but it did > not garner any responses. [1] > > In our small test deployments (160 HDs and OSDs across 20 machines) our > performance is quickly bounded by CPU and memory overhead. These are 2U > machines with 2x 6-core Nehalem; and running 8 OSDs consumed 25% of the > total CPU time. This was a cuttlefish deployment. > > This seems like a rather high CPU overhead. Particularly when we are > looking to hit density target of 10-15 4TB drives / U within 1.5 years. > Does anyone have suggestions for hitting this requirement? Are there ways > to reduce CPU and memory overhead per OSD? > > My one suggestion was to do some form of RAID to join multiple drives and > present them to a single OSD. A 2 drive RAID-0 would halve the OSD overhead > while doubling the failure rate and doubling the rebalance overhead. It is > not clear to me if that is better or not. > > [1] > http://lists.ceph.com/pipermail/ceph-users-ceph.com/2013-October/004833.html > > _______________________________________________ > ceph-users mailing list > ceph-users@lists.ceph.com > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > >
_______________________________________________ ceph-users mailing list ceph-users@lists.ceph.com http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com