Hacker News Reader

Comment by fragmede | Hacker News Reader

honestly I think btrfs isn't bloated enough for today's VM-enabled world. ext4 and xfs and hell, exfat haven't gone anywhere, and if those fulfill your needs, just use those. but if you need more advanced features that btrfs or zfs bring, those added features are quite welcome. imo, btrfs could use the benefits of being a cluster filesystem on top of everything it already does because having a VM be able to access a disk that is currently mounted by the host or another VM would useful. imagine if the disk exported to the VM could be mounted by another VM, either locally or remote simultaneously. arguably ceph fills this need, but having a btrfs-native solution for that would be useful.

View on HN

Running VMs (and database servers) on btrfs performs really bad so you have to disable CoW for them.

Otherwise you'll get situations where your 100GB VM image will use over a TB of physical disk space.

It's a shame really that this still isn't solved.

I don't think thin provisioning btrfs makes a lot of sense. Before disabling CoW I'd rather use a different filesystem.

Are you sure your TRIM is working and the VM disk image is compacting properly? It's working for me but not really great for fragmentation.

What’s the underlying issue? I used VMs with ZFS for storage for well over a decade with no issue.

ZFS performs much better than btrfs with the many small writes that VMs produce. Why exactly is a great question. Maybe it has to do with the optimizations around the ZIL, the temporary area where sync writes are accumulated before they are written to the long-term spot.

CoW won't necessarily make the VM image bloated. In fact, as I've foolishly found out, BTRFS can be quite useful for deduplicating very similar VMs at the block level, at the cost of needing to re-allocate new disk space on writes. In my VM archive, six 50 GiB virtual machines took up 52 GiB rather than 300 GiB and that was quite impressive.

Many downsides to CoW are also present with many common alternatives (i.e. thin LVM2 snapshots). Best to leave all of that off if you're using spinning rust or native compression features, though.

Checksum self healing on ZFS and BTRFS saved my data from janky custom NAS setups more times that I can count. Compression is also nice but the thing I like most is the possibility of creating many partition-like sub volumes without needing to allocate or manage space.

Reply on news.ycombinator.com