Btrfs slated for huge performance boost


Wow! The numbers looks it will make BTRFS much faster.

Can I say read and write will be much faster than EXT4 and NTFS?

you have to understand that Btrfs in 6.0 and 6.1 are a version 2.0
do they covers all errors ?

since 2016 , there is always change in kernels about Btfs it’s never ending
*** do not use with Raid 0 or 10 or 5 and 6 *** ( may be raid 1 can show strange results )

i prefer a rework for ext4 with io_uring for better perfomance

It is referring to write speeds.

see this about warnig Raid5/6

*Specifically there is also the RAID 5/6 page that does still carry the warning, "The parity RAID code has a specific issue with regard to data integrity. It should not be used for metadata. For data, it should be safe as long as a scrub is run immediately after any unclean shutdown.

Yes i saw that but then again for the average home user I’m not concerned about it. I think this warning is more for a different environment referring to server raid implementations where raid is more predominant. I find Raid is passe when it comes to most home user. It was a thing in the past for gaming but now a days with the hardware available i just don’t see it as something i care about.

I am not sure I understand what btrfs raid 5/6 being unsafe has to do with kernels commits that improve sequential wirte performance…

Btrfs raid 5/6 wasn’t recommended before this change and it won’t be recommended afterwards. It is simple enough to run btrfs on top of an MD device if you want both raid5/6 and btrfs.

Raid 5/6 would not be typically used for gaming. Are you thinking of raid 0?

Raid 5/6 is mostly used for redundancy to prevent data loss in the event of drive failure. I use it at home in several places, it is quite useful if you have a large volume of data. Of course, I don’t use btrfs raid 5 for the reasons mentioned above. All mine is zfs/md/lvm.

I didn’t bring it up and yes i was referring in general to Raid as in Raid 0 for gaming. I already saw the warning about raid 5/6 when reading the info. I was just saying i don’t think it’s a concern for the average home user as most wouldn’t be using raid although some do. You would know more about than i would as I’m no expert on btrfs or raid. It’s just my perception of Raid that it’s used more in other environments than an average home use.

Only write speeds not both read and write?
Anyway with SSD things will be much faster.

More details on the Btrfs update changelog for Kernel 6.1

  • Performances
  • Big refactoring

First I thought “wow, a good piece of work” but then I looked at the commit. It is really small, just changing 2 lines of code.

The benchmark is done with fio and the developer says: " To achieve the same or better performance with the existing code an io depth of 4 is required."

Together with the fact that this is for async writes only I am now wondering if normal users would benefit from this at all.