Right now, we create pools for images, compute, volumes, and radosgw creates a bunch, all are assigned to the default crush map.
from the Ceph side, In order to create a pool where we could separate it from another pool is to create a ruleset in the cursh map to isolate the devices, then the pools can be mapped to it. So once we create new ruleset, which pools would we map to it? do we create separate ones for each type of service? Just cinder? >From the fuel side, how would we distinguish between devices, and what do we assign them to? a pool (limited) or a ruleset (flexible, but then we need to code putting pools on them). Then to put them together, we are going to need a data driven crush tool writing, which I haven't seen. While here there we also need to think about how we might want to set replication between NodeGroups (which tend to logically map to a rack in a spine and leaf topology). I'd also loop in the puppet-openstack guys, because at some point I'd like to switch to the upstream module On Thu, Mar 19, 2015 at 7:12 PM, Rogon, Kamil <kamil.ro...@intel.com> wrote: > Hello, > > I want to initiate a discussion about different backend storage types for > Ceph. Now all types of drives (HDD, SAS, SSD) are treated the same way, so > the performance can vary widely. > > It would be good to detect SSD drives and create separate Ceph pool for > them. From the user perspective, it should be able to select pool when > scheduling an instance (scenario for high-IOPS needed VM, like database). > > Regards, > Kamil Rogon > ---------------------------------------------------------------------------- > --- > Intel Technology Poland sp. z o.o. > KRS 101882 > ul. Slowackiego 173 > 80-298 Gdansk > > > > __________________________________________________________________________ > OpenStack Development Mailing List (not for usage questions) > Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev > -- Andrew Mirantis Fuel community ambassador Ceph community __________________________________________________________________________ OpenStack Development Mailing List (not for usage questions) Unsubscribe: openstack-dev-requ...@lists.openstack.org?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev