Samsung Introduces New Linux File-System: F2FS

[johnc]

I never heard of anyone running into problems because his router wrote to much data to often to its flash storage.

Samsung is the world leader in flash technology.

Why would they spend money on this if they didn't think it was worthwhile?

[gururise]

Unlike lanyardfs, this one actually has a point. It'll maximize the life of the flash.

btrfs will not do that. Any generic fs will not do that.

Exactly. For those asking why we need YAFS (Yet another file system), this filesystem is highly tuned to the nuances of NAND flash (think Smartphone or Tablet FS). In addition to extending life, it is tuned/optimized to deliver consistent performance on NAND flash filesystems that are in widespread use today. Linux really has been lacking a filesystem tuned to the specifics of NAND flash, and this fills the gap. Thanks Samsung!

[blackout23]

I've killed several USB sticks. All any SSD is is just a RAID of USB sticks, to put it in common words.

It isn't. USB Sticks certainly don't use MLC NAND or even SLC NAND. They use cheaper Flash chips and inferior controllers.

[bug77]

I never heard of anyone running into problems because his router wrote to much data to often to its flash storage.

Think about memory in a smartphone and kids updating their FB accout every hour...
Plus, with the transition from 3x nm to 2x nm, flash P/E cycles took a nose dive. From 10k to 3-5k. Who knows how things will be in the future?

[curaga]

Well frankly, sometimes it's even worse, it might take 3-4 seconds to delete a few files worth of like 6GB, while on a window$ system it takes less than a second to do the same thing, often in a blink of an eye.

If your workload consists of often deleting big files, you should check the benchmarks and pick a FS more designed for that. Not to mention tune it for it.
The default FS is more tuned to small files and the usual day-to-day work.

(JFS and XFS are the usual recommendations for handling big files, but do benchmark.)

[curaga]

Trying to maximize flash lifetime is a worthless effort. Try writing a SSD to death withing 10 years.

http://www.xtremesystems.org/forums/showthread.php?271063-SSD-Write-Endurance-25nm-Vs-34nm

No one is ever going to write that much on an SSD on a daily basis that you exceed 1 PB host write.

A very common SSD (Crucial M4) has surpassed 750 TB host write. Spread it over 10 years (3650 Days) and you'd have
to write over 200 GB (the drive itself will have either 128 or 256 GB capacity) per day every day for 10 years and it will still function.
SSDs can take a lot of shit contrary to popular belief.

I've killed several USB sticks. All any SSD is is just a RAID of USB sticks, to put it in common words.

[blackout23]

This is for flash with no wear-leveling (like you might find in cheap embedded devices), not an SSD.

I never heard of anyone running into problems because his router wrote to much data to often to its flash storage.

[dnebdal]

Trying to maximize flash lifetime is a worthless effort. Try writing a SSD to death withing 10 years.

http://www.xtremesystems.org/forums/showthread.php?271063-SSD-Write-Endurance-25nm-Vs-34nm

No one is ever going to write that much on an SSD on a daily basis that you exceed 1 PB host write.

A very common SSD (Crucial M4) has surpassed 750 TB host write. Spread it over 10 years (3650 Days) and you'd have
to write over 200 GB (the drive itself will have either 128 or 256 GB capacity) per day every day for 10 years and it will still function.
SSDs can take a lot of shit contrary to popular belief.

This is for flash with no wear-leveling (like you might find in cheap embedded devices), not an SSD.

[droidhacker]

I've always found wondoze file deletes to be be incredibly slow. With Linux, you can delete a 500 GB file in a fraction of a second -- erase the inode and ignore the file, which is suddenly and thoroughly deallocated.

Now when you're dealing with MANY MANY MANY files, it can start to take a little longer, because it has to scan through, sort, and delete individually using a recursive algo. This could be more a limitation of the actual delete command than the filesystem itself.

Also; fat32 doesn't use any form of journal. If you want to compare performance of that, try ext2 vs fat or otherwise disable the journal.

[blackout23]

Unlike lanyardfs, this one actually has a point. It'll maximize the life of the flash.

btrfs will not do that. Any generic fs will not do that.

Trying to maximize flash lifetime is a worthless effort. Try writing a SSD to death withing 10 years.

http://www.xtremesystems.org/forums/showthread.php?271063-SSD-Write-Endurance-25nm-Vs-34nm

No one is ever going to write that much on an SSD on a daily basis that you exceed 1 PB host write.

A very common SSD (Crucial M4) has surpassed 750 TB host write. Spread it over 10 years (3650 Days) and you'd have
to write over 200 GB (the drive itself will have either 128 or 256 GB capacity) per day every day for 10 years and it will still function.
SSDs can take a lot of shit contrary to popular belief.