[sheepdog] [PATCH v3 6/9] sheep: introduce sparse objects

[Hitoshi Mitake]

At Thu, 27 Feb 2014 17:25:25 +0800,
Liu Yuan wrote:
> 
> On Sun, Feb 23, 2014 at 02:28:25PM +0900, Hitoshi Mitake wrote:
> > We don't need to care about a fragmentation when
> >  - the object is unlikely to be accessed sequentially, and
> >  - the object is read-only.
> > 
> > In that sense, we can make the objects sparse if they are not data
> > objects and writable ones.
> > 
> > This fixes the problem that sheepdog consumes many disk spaces for
> > deleted vdi objects if your filesystem supports FALLOC_FL_PUNCH_HOLE.
> > 
> 
> As far as I remember, this trim/untrim helpers are from old helpers that already
> to be proved buggy and hard to maintain. inode objects are considerably small
> compared to data objects, even it is deleted, it will be highly reused later.
> I don't see any real advantage for this patch set. You assume users generate
> huge vdis and then delete it, and then never re-create vdis. I don't think this
> is practical. Besides, this patch looks pre-mature optimization to me because we
> have noticed that total number of VDIS are graduately increasing, so making this
> sparse optimization in vain because there is always very limited zero-ed inodes.
> 
> Either way, I think we don't this patch.

The sparse object aims to reduce disk consumption of ledger objects,
not inode. The patch marks the inode objects as sparse ones but it is
just side effect.

Thanks,
Hitoshi