I'm not sure about the implications of IB, but Ceph certainly has striping capability. The following section of the architecture document explains it. See http://ceph.com/docs/master/architecture/#how-ceph-clients-stripe-data
On Fri, Mar 1, 2013 at 9:28 AM, Gandalf Corvotempesta <[email protected]> wrote: > Let's assume a cluster with IB OSD network and 2GB public network (2 > bonded nic for each storage server) > > Public network is used by RGW, MON, MDS and server access. > With replica set to 3, is ceph smart enough to route read requests for > the same object to to multiple servers archieving something like 3x2GB > of aggregated bandwidth? > _______________________________________________ > ceph-users mailing list > [email protected] > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com -- John Wilkins Senior Technical Writer Intank [email protected] (415) 425-9599 _______________________________________________ ceph-users mailing list [email protected] http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
