On 10 September 2013 15:15, Diego Parrilla Santamaría < diego.parrilla.santamaria at gmail.com> wrote: > You are describing the problems of using a shared filesystem backend for > cinder, instead of using a driver with direct connection at block-device > level. > > This is how it is implemented right now in Grizzly. You are right about block-device level but this is not where it goes. Take a look at https://blueprints.launchpad.net/cinder/+spec/qemu-assisted-snapshots > It has improved a lot in the last 18 months or so, specially if you want > to use as shared storage for your VMs. > > Seems the snapshotting feature is on the way: > > https://github.com/openstack/cinder/blob/master/cinder/volume/drivers/glusterfs.py > > Exactly, using qcow. > But the killer feature is the direct access from QEMU to Gluster using > libgfapi. It seems it has been added in Havana and it's in master branch > since mid August: > https://review.openstack.org/#/c/39498/ > > If I had to consider a scalable storage solution for an Openstack > deployment for the next 10 years, I would consider Gluster. > Just out of curiosity. Have you tested any other Cinder backends? regards -- Maciej Gałkiewicz Shelly Cloud Sp. z o. o., Sysadmin http://shellycloud.com/, macias at shellycloud.com KRS: 0000440358 REGON: 101504426 -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.openstack.org/pipermail/openstack/attachments/20130910/68c65fe4/attachment.html>