On 12-03-15 13:00, Lindsay Mathieson wrote: > On Thu, 12 Mar 2015 12:49:51 PM Vieresjoki, Juha wrote: >> But there's really no point, block storage is the only viable option for >> virtual machines performance-wise. With images you're dealing with multiple >> filesystem layers on top of the actual block devices, plus Ceph as block >> storage supports pretty much everything that qcow2 as a format does. > > > Not much difference that I've noticed on the small system I run, in fact cepfs > seems to do some caching which speeds up things considerably in benchmarks. > Not a huge difference in actual app performance that I've noticed. > > Delete/Restore of snapshots is a lot quicker with qcow2 and for somereasons > saving memory state is orders of magnitude quicker with qcow2. Don't, simply don't run your VMs with QCOW2 over CephFS. You bring in so much more complexity by adding a Posix Filesystem and the Ceph MDS. Although CephFS might look faster due to the kernel caching, it's not safer! Use RBD with Qemu, it keeps it all in userspace, a lot better then going through all the kernelspace things. Wido > _______________________________________________ > ceph-users mailing list > ceph-users@xxxxxxxxxxxxxx > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |