Btrfs/EXT4/XFS/F2FS RAID 0/1/5/6/10 Linux Benchmarks On Four SSDs

[phoronix]

Phoronix: Btrfs/EXT4/XFS/F2FS RAID 0/1/5/6/10 Linux Benchmarks On Four SSDs

Following the recent Btrfs RAID: Native vs. Mdadm comparison, the dual-HDD Btrfs RAID benchmarks, and four-SSD RAID 0/1/5/6/10 Btrfs benchmarks are RAID Linux benchmarks on these four Intel SATA 3.0 solid state drives using other file-systems -- including EXT4, XFS, and Btrfs with Linux 3.18.

http://www.phoronix.com/vr.php?view=21124

[thebear]

An interesting read. Thanks!

I wonder how e.g. ext4 would fare with:

mkfs.ext4 -O ^has_journal

and mount options:

noatime,nodiratime,noacl,discard

in e.g. the compile tests, obviously this is for temporary work space (not for you home directory etc..)

oh and by the way, would 4 SSD's scale even with newer, bigger and faster ones (say 4x 500GB modern SSDs) or
could one expect a bottleneck?

[thelongdivider]

Thanks for this!

This article is a very interesting look. I've been fascinated with BTRFS and all the room for expansion it seems to have. It looks like native btrfs raid 5 ofver 4 HDDs wasn't the worst call (even starting it at kernel 3.10).

[yakman2020]

First off, really nice work, very informative, and very helpful. Makes me want to look more closely at Btfs.

But... I'm wondering how the FSs and their options compare in terms of maturity and resiliency. Speed does no good if you lose data from time to time due to bugs in the FS code. Or are all of the options pretty much equal in that regard?

And that works both ways. If you can be sure you won't lose power (for example, a robust UPS) it might be interesting to see how the FS perf compares tuned to max performance with logs in memory, etc, rather than "normal" levels of failure tolerance.

[xxmitsu]

It would be really useful the number of max tps/IOPS for each setup.. and also some benchmark with SSDs at 80% of full capacity.
One question: is it TRIM supported in any of those setups?

[profoundWHALE]

I'm impressed at how far btrfs has come. I wonder if we'll start seeing it replacing EXT4 in a few years.

@thebear
I don't think he would test with those options unless he did a separate benchmark that compared performance of various mounting options.

@xxmitsu
If you really need specific SSD performance metrics, you might as well take a look at some stuff that Anandtech has

[Michael]

It would be really useful the number of max tps/IOPS for each setup.. and also some benchmark with SSDs at 80% of full capacity.
One question: is it TRIM supported in any of those setups?

TRIM is supposed to work with the native Btrfs RAID.

[rtreffer]

degraded performance?

I would really like to know how well the performance degrades if disks are missing.
RAIDs are often used to survive a hardware failure. But this will only work out if the performance stays within reasonable ranges.

I've seen way too many system die because the RAID didn't perform as expected - either to disk failure or due to latent disk/controller errors.

[cjcox]

I'm impressed at how far btrfs has come. I wonder if we'll start seeing it replacing EXT4 in a few years.

Uh... done deal already. Unless you are trapped in the whole USA, Linux === Red Hat thing... in which case I think it's best to stop saying Linux or Linux distro and just say Red Hat (but beware of the trademark police).

[stiiixy]

Uh... done deal already. Unless you are trapped in the whole USA, Linux === Red Hat thing... in which case I think it's best to stop saying Linux or Linux distro and just say Red Hat (but beware of the trademark police).

At least they are open to linux, and have mega-bomb dollar company's developing, building and supporting it on real systems. Can't say the same for some country's.