Btrfs Has Many Nice Improvements, Better Performance With Linux 5.11

[phoronix]

Phoronix: Btrfs Has Many Nice Improvements, Better Performance With Linux 5.11

The first set of major file-system material submitted for the newly opened Linux 5.11 merge window are the Btrfs updates...

http://www.phoronix.com/scan.php?page=news_item&px=Btrfs-Linux-5.11-Improvements

[Tori]

Cool, what HW will you use for the tests? Perhaps the new WD Black NVMe SSD?

[Hans Bull]

ignoredatacums

Please...

[Templar82]

There has been pretty consistent performance increases lately with BTRFS, It will be interesting to see some benchmarks.

[EarthMind]

There seems to have been a lot of interesting improvements after Red hat decided to say it's a dead project. I wonder if they've already realized they made a poor judgment. It's good to see that Red hat alone doesn't rule the Linux world.

[chilek]

There seems to have been a lot of interesting improvements after Red hat decided to say it's a dead project. I wonder if they've already realized they made a poor judgment. It's good to see that Red hat alone doesn't rule the Linux world.

You've probably missed some interesting news. BTRFS is default filesystem for root in Fedora 33! That means that if it doesn't change for a few editions it will have a big chance to migrate to RHEL 9 (probably).

[DonnieWest]

I'd be super curious to see how Btrfs and ZFS compare nowadays. Admittedly, I've not looked into either (I've stuck with EXT4) but maybe it's finally time to check them out

[You-]

As always the main issue with file systems is always stability.

There has been near silence on that front since the release of Fedora 33, so that is a very positive sign.

AFAIk Facebook (normally an evil company) has put in a lot of work and hired many developers to iron out any issues and make it something they can "rely" upon, so we all benefit.

A use case and performance comparison with OpenZAFS would be good to read if it was by someone who uses the features (at the same time without being too technical).

I have read that the Copy on Write features can be bad for VMs and databases but I have never understood why and if there are protections to turn off CoW in those situations.

[cynic]

I have read that the Copy on Write features can be bad for VMs and databases but I have never understood why and if there are protections to turn off CoW in those situations.

the reason is that COW, especially in that use-cases, creates a lot of fragmentation

[Volta]

I have read that the Copy on Write features can be bad for VMs and databases but I have never understood why and if there are protections to turn off CoW in those situations.

It seems:

Libvirt will by default set chattr +C on the enclosing directory for VM images for Cockpit, virt-manager, GNOME Boxes, and virt-install. So users won’t need to worry about this.

https://fedoramagazine.org/btrfs-coming-to-fedora-33/