Btrfs vs. EXT4 vs. XFS vs. F2FS On Linux 3.10

[wargames]

ext4 forever

ext4 forever! Seriously, do we really need so many filesystems ? Benchmark after benchmark, I fail to see any advantage of all those filesystems compared to ext4.

[Calinou]

For long term storage these people will still be using one of the "slow" filesystems that actually protect their data.

It's not like ext4 isn't stable at all. The only upside of btrfs about stability is snapshots; it is usually less stable than ext4.

Everyone else should pick the filesystem with the best data integrity and resiliance. You will likely never notice loss of "speed" what with continuous improvements in hardware, caching and prefetch of data.

Everyone else should encrypt their multiplayer game packets in AES, too, because you don't want other players to know your location on the map, but we can afford our ping times going from 50 ms to 150 ms. Amirite?

Ontopic : it's quite funny, but even on a SSD, ext4 tends to be better than F2FS. I'm using it on a SSD and it works fine.

[Siuoq]

what about HUMMER, reFS and tux3?

I think if you are very bored you should try to make a grand filesystem speed comparison.

[Sergio]

Still don't get it...

I still fail to get the point of these apples-vs-oranges benchmarks...

[Ericg]

ext4 forever! Seriously, do we really need so many filesystems ? Benchmark after benchmark, I fail to see any advantage of all those filesystems compared to ext4.

Ext4's architecture and design is 20years old, its boilerplate code atop ext1. Thats about as much of a "hackjob" as you can get. So tell me.. do you like the thought of your entire filesystem being a hack job?

Besides, every filesystem has its own place and own goal. F2F2 is designed or flash media like phones and tablets. Btrfs is a pretty good general filesystem, especially if have limited space (compression) or multiple disks (built in volume management), it also has functions specifically built in for ssd's which is very relevent in the future, something that ext4 doesn't have.

XFS, JFS, Reiser, those..I honestly couldn't tell you what they are good for. From what I hear, XFS is very prone to complete failures and even one dirty shutdown can hose the entire system. JFS was an IBM filesystem from forever ago, Reiser was a proof-of-concept filesystem who's reputation got trashed after its head developer (and the person it was named after) murdered his wife.

What I am looking forward to though is Tux3, It will be very interesting to see the results of Tux3 vs Btrfs in the next few years.

--A happy btrfs user--

[deanjo]

From what I hear, XFS is very prone to complete failures and even one dirty shutdown can hose the entire system. -

A possibility, yes, prone, I wouldn't says so. In 15 years I have yet to actually come across any corruption or failures using XFS and trust me, there have been quite a few dirty shutdowns. I've actually had more issues with data integrity with EXT 3 then with XFS.

[dalingrin]

Ext4's architecture and design is 20years old, its boilerplate code atop ext1. Thats about as much of a "hackjob" as you can get. So tell me.. do you like the thought of your entire filesystem being a hack job?

Besides, every filesystem has its own place and own goal. F2F2 is designed or flash media like phones and tablets. Btrfs is a pretty good general filesystem, especially if have limited space (compression) or multiple disks (built in volume management), it also has functions specifically built in for ssd's which is very relevent in the future, something that ext4 doesn't have.

XFS, JFS, Reiser, those..I honestly couldn't tell you what they are good for. From what I hear, XFS is very prone to complete failures and even one dirty shutdown can hose the entire system. JFS was an IBM filesystem from forever ago, Reiser was a proof-of-concept filesystem who's reputation got trashed after its head developer (and the person it was named after) murdered his wife.

What I am looking forward to though is Tux3, It will be very interesting to see the results of Tux3 vs Btrfs in the next few years.

--A happy btrfs user--

I'm really happy with btrfs also. I know SSD's are all the rage these days but I also have older hard disks for storage and BTRFS w/LZO makes a huge difference on rotational disks. I can't tell a big difference with LZO on my SSD but its great on my other disks. BTRFS's native snapshot ability sounds great too but I've not used it yet.

[stan]

BTRFS is slow as ****

So let me see... BTRFS is slower than EXT4 pretty much everywhere, sometimes MUCH slower. All the purported features of BTRFS that are supposed to make BTRFS better are pie-in-the-sky as far as I can tell. Has anyone actually used the BTRFS snapshot thing on their Linux Desktop? Is there even a GUI for it, something that can come close to Apple's Time Machine? From a Linux Desktop user's perspective, BTRFS is worthless right now, and development resources going into it are a waste.

As for compression, I'm sorry, but I don't want to waste CPU and get latency because the underlying filesystem sucks at arranging data on the hard disk.

[smitty3268]

So let me see... BTRFS is slower than EXT4 pretty much everywhere, sometimes MUCH slower.

I don't know, it looked pretty competitive in most tests. There were just some where it was way slower.

All the purported features of BTRFS that are supposed to make BTRFS better are pie-in-the-sky as far as I can tell. Has anyone actually used the BTRFS snapshot thing on their Linux Desktop? Is there even a GUI for it, something that can come close to Apple's Time Machine?

Yes, it works well. No, there's no fancy GUI, but the underlying tech works fine.

As for compression, I'm sorry, but I don't want to waste CPU and get latency because the underlying filesystem sucks at arranging data on the hard disk.

Compression results in LESS latency, because the CPU can decompress things faster than IO can read data from a disk.

[IsacDaavid]

True, wasn't thinking about the fact that he just does zero-filled files instead of random-filled

Off-topic: From a merely theoretical standpoint, is there any benefit in (losslessly) compressing truly random data?