BTRFS on Debian Stable, your experiences?

If none of the more specific forums is the right place to ask

Re: BTRFS on Debian Stable, your experiences?

Postby bester69 » 2020-08-24 15:13

Head_on_a_Stick wrote:Fedora have just announced they're moving to Btrfs as the default filesystem for v33:

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

Great News to know BTRFS goes on fordward... there was a past time there were some doubts about btrfs future consolidation
bester69 wrote:You wont change my mind when I know Im right, Im not an ...
User avatar
bester69
 
Posts: 1739
Joined: 2015-04-02 13:15

Re: BTRFS on Debian Stable, your experiences?

Postby LE_746F6D617A7A69 » 2020-08-24 20:57

I'd like to see an unbiased comparison of BTRFS RAID 1/10 vs mdadm RAID 1/10 - I've made the tests for myself, but nobody would believe me - too many fanboys ...
Bill Gates: "(...) In my case, I went to the garbage cans at the Computer Science Center and I fished out listings of their operating system."
The_full_story and Nothing_have_changed
LE_746F6D617A7A69
 
Posts: 389
Joined: 2020-05-03 14:16

Re: BTRFS on Debian Stable, your experiences?

Postby Mr. Lumbergh » 2020-08-26 19:54

LE_746F6D617A7A69 wrote:I'd like to see an unbiased comparison of BTRFS RAID 1/10 vs mdadm RAID 1/10 - I've made the tests for myself, but nobody would believe me - too many fanboys ...

You noticed that too, eh? :D
Mr. Lumbergh
 
Posts: 97
Joined: 2019-08-02 04:28

Re: BTRFS on Debian Stable, your experiences?

Postby Deb-fan » 2020-10-09 23:43

How does btrfs do on things like SSD's with compression enabled ? Is there any notable speed gains during transfer, disk i/o operations ? Btrfs is certainly on the things I want to mess with list though. Don't always go with the version chasers in everything tech but yeppers, would be wanting latest-greatest in something like the btrfs filesystem. :)
Most powerful FREE tech-support tool on the planet * HERE. *
Deb-fan
 
Posts: 939
Joined: 2012-08-14 12:27

Re: BTRFS on Debian Stable, your experiences?

Postby pylkko » 2020-10-10 11:24

LE_746F6D617A7A69 wrote: I've made the tests for myself, but nobody would believe me - too many fanboys ...
But if you produced such test results, what is the reason that you cannot publish them? Or have you? Because you cannot really expect people to believe otherwise. Or at least, it would draw much more interest towards the issue.
Deb-fan wrote:How does btrfs do on things like SSD's with compression enabled ? Is there any notable speed gains during transfer, disk i/o operations ? Btrfs is certainly on the things I want to mess with list though. Don't always go with the version chasers in everything tech but yeppers, would be wanting latest-greatest in something like the btrfs filesystem. :)
What do you mean exactly? Are you asking if the performance is better or worse than something else X? What do you want to compare to will decide if there is an already-existing benchmark out there on it. It is probably not possible to give a complete answer as it will depend on the hardware in question. When I install on btrfs I use compression and do it on an SSD. On the fly compression of course consumes compute time on the processor, but it will save space, and as discussed earlier, can increase throughput in cases where the compute operation is quicker than the transfer and what you are moving around compresses well.
User avatar
pylkko
 
Posts: 1791
Joined: 2014-11-06 19:02

Re: BTRFS on Debian Stable, your experiences?

Postby Deb-fan » 2020-10-11 14:01

^ Yeah was pretty much it, was wanting feedback from someone using btrfs on an SSD. Of course there's no easy answer I guess, like everything it must depend. A lot of really tech-literate people have spent considerable time working on btrfs so it's bound to be interesting stuff. :)
Most powerful FREE tech-support tool on the planet * HERE. *
Deb-fan
 
Posts: 939
Joined: 2012-08-14 12:27

Re: BTRFS on Debian Stable, your experiences?

Postby Head_on_a_Stick » 2020-10-11 14:24

FWIW, acording to Phoronix btrfs is slower than xfs or f2fs on NVMe drives under kernel 5.8: https://www.phoronix.com/scan.php?page= ... tems&num=1

I've noticed that btrfs is *slightly* slower than xfs or f2fs on my M2 NVMe drive but it's not a big difference and anyway I use btrfs for the features (snapshots, subvolumes, automatic error-correction and the damage resistance conferred by the copy-on-write paradigm) rather than the performance.
Black Lives Matter

Debian buster-backports ISO image: for new hardware support
User avatar
Head_on_a_Stick
 
Posts: 12629
Joined: 2014-06-01 17:46
Location: /dev/chair

Re: BTRFS on Debian Stable, your experiences?

Postby Deb-fan » 2020-10-11 14:57

^ Thanks and that's about as recent as it gets 5.8, so is definitely appreciated and pointing out the full feature set too. Though this on-the-fly de/compression thing has had me fascinated for a long time. Along with the parallel/multicore tools like pigz etc. Btrfs is clearly something designed mostly for enterprise but still interesting stuff.
Most powerful FREE tech-support tool on the planet * HERE. *
Deb-fan
 
Posts: 939
Joined: 2012-08-14 12:27

Re: BTRFS on Debian Stable, your experiences?

Postby LE_746F6D617A7A69 » 2020-10-11 17:47

pylkko wrote:
LE_746F6D617A7A69 wrote: I've made the tests for myself, but nobody would believe me - too many fanboys ...
But if you produced such test results, what is the reason that you cannot publish them? Or have you? Because you cannot really expect people to believe otherwise. Or at least, it would draw much more interest towards the issue.
Currently I can't repeat the tests, because that would require to reformat My drives - and I'm not going to do that just to to satisfy Your curiosity.
However I remember very well why I've decided to go with md arrays + ext4 and not with BTRFS. There were 2 main reasons:
1. All BTRFS Raid levels except level 1 are useless, in terms of performance (compared to md arrays: nearly 2 times slower)
2. Unlike md arrays, BTRFS increases access time instead of reducing it. F.e. on BTRFS Raid10 I've got ~13..14ms while on md Raid10 I have ~10ms (single disk: 12ms) - this is a show stopper, because it kills random access performance.

The test was performed on ridiculously cheap array of 4x WD RED RE4 500GB HDDs (single-platter 500GB HDDs are the most reliable HDDs on the market today). Programs used for testing: iozone, seeker.c and GNOME Disks.

---------------------
I'm not using SSDs to keep important data - the 2 SSDs I have installed in my PC are used only for virtual machines.
The md Raid10 array made of 4 HDDs is nearly as fast as a single SSD - but it does not suffer from write amplification, and it offers steady performance through the whole lifetime of the devices (unlike SSD). Besides, I can turn off the PC for a month, without risking of loosing 3% of random data - which is "acceptable" for SSDs, accordingly to JEDEC specifications. (that 3% is for TLC, and somehow they forgot to update the specs for QLC :lol: )
Bill Gates: "(...) In my case, I went to the garbage cans at the Computer Science Center and I fished out listings of their operating system."
The_full_story and Nothing_have_changed
LE_746F6D617A7A69
 
Posts: 389
Joined: 2020-05-03 14:16

Re: BTRFS on Debian Stable, your experiences?

Postby Deb-fan » 2020-10-11 21:33

^ Interesting insights, thanks for sharing. The this, that vs btrfs thing, just to be diplomatic about things, good news is nobody has to be wrong, we can all use any filesystem(s) we like, for any reason wanted and it's still all good. :)

Ps, besides everybody knows ntfs is the absolute best filesystem on earth. :P
Most powerful FREE tech-support tool on the planet * HERE. *
Deb-fan
 
Posts: 939
Joined: 2012-08-14 12:27

Re: BTRFS on Debian Stable, your experiences?

Postby LE_746F6D617A7A69 » 2020-10-11 21:55

Deb-fan wrote:Ps, besides everybody knows ntfs is the absolute best filesystem on earth. :P

http://forums.debian.net/viewtopic.php?f=3&t=146274&start=15#p722183
:lol:
Bill Gates: "(...) In my case, I went to the garbage cans at the Computer Science Center and I fished out listings of their operating system."
The_full_story and Nothing_have_changed
LE_746F6D617A7A69
 
Posts: 389
Joined: 2020-05-03 14:16

Re: BTRFS on Debian Stable, your experiences?

Postby Deb-fan » 2020-10-12 00:18

^ LMAO ... though most of that went right over my head, you clearly know your stuff in tech fellow nixer. Am not intimately familiar with the workings of any filesystem. Just glad there's a lot better tech-literate people to develop them, so hopefully will get around to trying them all to some extent.
Most powerful FREE tech-support tool on the planet * HERE. *
Deb-fan
 
Posts: 939
Joined: 2012-08-14 12:27

Re: BTRFS on Debian Stable, your experiences?

Postby pylkko » 2020-10-12 13:29

LE_746F6D617A7A69 wrote:
pylkko wrote:
LE_746F6D617A7A69 wrote: I've made the tests for myself, but nobody would believe me - too many fanboys ...
But if you produced such test results, what is the reason that you cannot publish them? Or have you? Because you cannot really expect people to believe otherwise. Or at least, it would draw much more interest towards the issue.
Currently I can't repeat the tests, because that would require to reformat My drives - and I'm not going to do that just to to satisfy Your curiosity.
Then maybe don't complain that nobody will believe you. You do not need to repeat the tests, you only need to describe them so clearly that others could repeat or find somebody else that did and wasn't afraid to publish

some interesting links
benchmarks
https://www.phoronix.com/scan.php?page= ... 4-hdd-raid
guy simulates errors on drives with btrfs, zfs and other configs
http://www.unixsheikh.com/articles/batt ... grity.html
User avatar
pylkko
 
Posts: 1791
Joined: 2014-11-06 19:02

Re: BTRFS on Debian Stable, your experiences?

Postby LE_746F6D617A7A69 » 2020-10-12 14:54

That benchmark on Phoronix is worth shit, because Mr Larabel has no idea how to setup md Raid10 and how to configure Ext4 for use with the raid array.
man mkfs.ext4 wrote:stride=stride-size
Configure the filesystem for a RAID array with stride-size filesystem blocks. This is the number of blocks read or written to disk before moving to the next disk, which is sometimes referred to as the chunk size. This mostly affects placement of filesystem metadata like bitmaps at mke2fs time to avoid placing them on a single disk, which can hurt performance. It may also be used by the block allocator.

stripe_width=stripe-width
Configure the filesystem for a RAID array with stripe-width filesystem blocks per stripe. This is typically stride-size * N, where N is the number of data-bearing disks in the RAID (e.g. for RAID 5 there is one parity disk, so N will be the number of disks in the array minus 1). This allows the block allocator to prevent read-modify-write of the parity in a RAID stripe if possible when the data is written.

Besides, different stripe sizes are optimal for different workloads and for different device types - both md and ext4 allows to freely select the chunk size, and in btrfs there's no such option.

pylkko wrote:guy simulates errors on drives with btrfs, zfs and other configs
http://www.unixsheikh.com/articles/batt ... grity.html

This guy made many mistakes during testing - fortunately he have already noticed some of them .... ;)
But seriously, I'm using md arrays for over 15 years - they have survived hundreds of power outages and many disk failures - no single problem.
Bill Gates: "(...) In my case, I went to the garbage cans at the Computer Science Center and I fished out listings of their operating system."
The_full_story and Nothing_have_changed
LE_746F6D617A7A69
 
Posts: 389
Joined: 2020-05-03 14:16

Re: BTRFS on Debian Stable, your experiences?

Postby pylkko » 2020-10-12 17:59

LE_746F6D617A7A69 wrote:That benchmark on Phoronix is worth shit, because Mr Larabel has no idea how to setup md Raid10 and how to configure Ext4 for use with the raid array.

Yes, if someone knew how to design a better, more fair and meanigful benchmark, people would not ignore it, for sure as shit! Right now, this is the best we have.

This guy made many mistakes during testing - fortunately he have already noticed some of them .... ;).


Yes, it is very well written. Especially that he admits his shortcomings and accepts criticism (as opposed to calling all other views flawed while not exposing his own) and even works with the posters to improve and redo stuff. This is really good stuff, and I haven't found any other similar post yet. It is unique in its kind.
User avatar
pylkko
 
Posts: 1791
Joined: 2014-11-06 19:02

PreviousNext

Return to General Questions

Who is online

Users browsing this forum: esalvesen and 16 guests

fashionable