Linux 5.14 SSD Benchmarks With Btrfs vs. EXT4 vs. F2FS vs. XFS

[coder]

Don't imply that disks internal csum is of value here. Most have them, but reality tells us it doesn't matter. There are several failures that can happen anyway.

Some of them are :

• Not honouring barriers
• Lost cache on bus, disk or host bus resets
• Fimware bugs
• Powersave bugs
• Bad controllers
• Bad USB-SATA bridges
• ...

Even with enterprise hardware these things happen. Even on enterprise raid controllers these things happen and those controllers don't have methods to handle corrupt data from individual disks.

Most of these bugs can screw BTRFS, too, depending on just how bad they hit you!

And BTRFS RAID is no magic bullet:

https://btrfs.wiki.kernel.org/index.php/RAID56

[sinepgib]

Last I checked, SSD wear is a non-issue on any recent SSDs, unless perhaps you're using the cheapest QLC you could find.

How come? Isn't it still a consequence of flash technology that it wears on erasure?

[S.Pam]

How come? Isn't it still a consequence of flash technology that it wears on erasure?

Yes you are correct. But large drives are better at wear levelling and have much higher TBW.

Endurance is approximately:

SLC=100k writes
MLC=10k writes
TLC=3k writes
QLC=1K writes

[Azrael5]

It looks like F2Fs be the best solution for SSD and I assume USB as well.

[zoner]

Response to Eumaios

My comment were not directed at Michael but that comment was directed to all of the comments about others about why their computer was slow.

[Zeioth]

How come that F2FS improved so much? It's a kernel 5.14 thing?

[S.Pam]

So .......btrfs on hard drive for home/personal server use cases should be fine right?

Yes.

If your disk is /dev/sdb1 then use:

# mkfs.btrfs -R free-space-tree -L my-btrfs-disk /dev/sdb1

To mount:

# mount LABEL=my-btrfs-disk /mnt/my-btrfs-root -o noatime,subvolid=5
# btrfs subvolume create /mnt/my-btrfs-root/@home
# mount LABEL=my-btrfs-disk /home -o noatime,subvol=@home

Some other useful commands are

Lists all known Btrfs filesystems on the system:

# btrfs filesystem show
Label: 'btrfs-root' uuid: 446d32cb-a6da-45f0-9246-1483ad3420e0
Total devices 1 FS bytes used 89.60GiB
devid 1 size 229.47GiB used 99.06GiB path /dev/sda3

Label: 'boot' uuid: b1ae03e7-e6c2-4efe-90ec-57a54e296e2e
Total devices 1 FS bytes used 35.64MiB
devid 1 size 1.00GiB used 342.38MiB path /dev/sda2

Label: 'usb-backup' uuid: df68a30d-d26e-4b9c-9606-a130e66ce63d
Total devices 1 FS bytes used 685.68GiB
devid 1 size 927.51GiB used 721.02GiB path /dev/sde1

Show disk space usage and allocation (instead of using df tool):

 # btrfs filesystem usage -T /mnt/rootvol/
Overall:
Device size: 229.47GiB
Device allocated: 99.06GiB
Device unallocated: 130.41GiB
Device missing: 0.00B
Used: 92.03GiB
Free (estimated): 134.23GiB (min: 69.03GiB)
Free (statfs, df): 134.23GiB
Data ratio: 1.00
Metadata ratio: 2.00
Global reserve: 382.42MiB (used: 0.00B)
Multiple profiles: no

Data,single: Size:91.00GiB, Used:87.18GiB (95.80%)
/dev/sda3 91.00GiB

Metadata,DUP: Size:4.00GiB, Used:2.42GiB (60.63%)
/dev/sda3 8.00GiB

System,DUP: Size:32.00MiB, Used:16.00KiB (0.05%)
/dev/sda3 64.00MiB

Unallocated:
/dev/sda3 130.41GiB

Show logged error count

# btrfs device stats /mnt/rootvol/
[/dev/sda3].write_io_errs 0
[/dev/sda3].read_io_errs 0
[/dev/sda3].flush_io_errs 0
[/dev/sda3].corruption_errs 0
[/dev/sda3].generation_errs 0

Scrub your disk regularly (monthly or so is enough)

#btrfs scrub start /mnt/rootvol/
scrub started on /mnt/rootvol/, fsid 446d32cb-a6da-45f0-9246-1483ad3420e0 (pid=8175)

# btrfs scrub status /mnt/rootvol/
UUID: 446d32cb-a6da-45f0-9246-1483ad3420e0
Scrub started: Sun Aug 29 13:40:26 2021
Status: running
Duration: 0:00:45
Time left: 0:02:20
ETA: Sun Aug 29 13:43:32 2021
Total to scrub: 92.03GiB
Bytes scrubbed: 22.36GiB (24.30%)
Rate: 508.87MiB/s
Error summary: no errors found

Enable compression on some files/folders (only newly written data will be compressed)

# btrfs property set <dir-or-file> compression zstd

One-time compression of existing files.

# btrfs filesystem defragment -czstd -vr <dir-or-file>

Use compress mount option

# mount -o compress=zstd
# mount -o compress=zstd:5
# mount -o compress=lzo
# mount -o compress (this is zlib)
# mount -o compress-force=zstd
# mount -o compress-force=zstd:15

Reference: https://wiki.tnonline.net/w/Category:Btrfs

[F.Ultra]

he means database takes care of data integrity even without any filesystem(on raw block device). that's database's job

Only if enabled though and AFAIK both MySQL/MariaDB and PostgreSQL have it disabled as default.

[gadnet]

Can I haz the same tests, but with btrfs nodatacow added?

what's the use of btrfs if you disable snapshots and checksumming, use xfs/ext4 then
it would be better to test all FS with also ZFS , with noatime and in a raid 1 config as you allways want to have at least raid1 security and who run atime in a cow file system ?

best regards.

[S.Pam]

what's the use of btrfs if you disable snapshots and checksumming, use xfs/ext4 then
it would be better to test all FS with also ZFS , with noatime and in a raid 1 config as you allways want to have at least raid1 security and who run atime in a cow file system ?

best regards.

You still get snapshots and all the other things with btrfs. Even reflinks works with nodatacow.