Re: Live Fille System Backup
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Tue, May 09, 2017 at 08:08:48AM -0400, Greg Wooledge wrote:
[...]
> So, basically the same as LVM snapshots, which are independent of the
> type of file system that sits on the LVM logical volume.
To be fair, btrfs knows about the file system, whereas LVM does COW
on the underlying "dumb" block device. Thus btrfs is capable of
squeezing quite a bit out in terms of better performance/lower disk
usage.
For some workloads it matters, for others not (the use case I posited,
freezing a snapshot while rsyncing, might tend to fall on the "not"
side *if* you are a bit careful and *if* your write load isn't insane).
> But btrfs is an edge with more blood on it, so people are naturally
> drawn to it. It's also more complicated. So it's *got* to be better!!!
>
> > I don't know what an atomic cow is (maybe some poor bovine living near
> > Chernobyl) but it sounds interesting.
>
> C.O.W. in this context is usually "copy on write".
>
> Atomic means that it's like a database transaction: the entire operation
> either succeeds, or fails. There's no half-way state.
>
> > I wonder if Btrfs has any notable drawbacks as a file system.
>
> Some of the blood on btrfs is starting to dry, so the blood-seekers
> are all flocking to ZFS now. It's got a whole new TIER of buzzwords
> and complexity! It's GOT to be the best!!!
:-)
ISTR Android has btrfs as default these days? (careful, I didn't check,
and my brain has one fake news source embedded somewhere I couldn't
yet stomp out. Might be wrong).
cheers
- -- t
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)
iEYEARECAAYFAlkRtdAACgkQBcgs9XrR2kZ7fgCcD/B1+IavQMGeEJ51Y4L6uoPe
WaMAnR/seOj+evs/hk7Wb3bJ/TtYBUP2
=3Jdh
-----END PGP SIGNATURE-----
Reply to: