On Thu, Jan 06, 2011 at 05:50:42PM +0100, Michael Monnerie wrote: > On Donnerstag, 6. Januar 2011 Dave Chinner wrote: > > Realistically, we have > > to design for how we expect devices to behave in 2-3 years time, not > > waste time trying to optimise for fundamentally broken devices that > > nobody will be using in 2-3 years time... > > I see, thanks for the explanation. I hope that there will be a solution > that allows us to FITRIM on Saturday "in the background", so the thin > provisioned VMs are smaller and faster to back up. That's my immediate > wish, as we currently don't use SSDs. That's what you can do with my FITRIM patches. Right now the only discard support known to me is my impementation that requires XFS as backing store for the images. And a hole punch of an existing hole is very cheap in XFS, so trimming things multiple times is not an issue. _______________________________________________ xfs mailing list xfs@xxxxxxxxxxx http://oss.sgi.com/mailman/listinfo/xfs