Following the recent announcement that UBIFS is nearly production ready, it was asked how UBIFS compares to LogFS. LogFS author Jörn Engel suggested, "both share similar design goals. Biggest difference is that ubifs works on top of ubi and depends on ubi support, while logfs works on plain mtd (or block devices) and does everything itself. Code size difference is huge. Ubi weighs some 11kloc, ubifs some 30, logfs some 8." He continued:
"Ubi scales linearly, as it does a large scan at init time. It is still reasonably fast, as it reads just a few bytes worth of header per block. Logfs mounts in O(1) but will currently become mindbogglingly slow when the filesystem nears 100% full and write are purely random. Not that any other flash filesystem would perform well under these conditions - it is the known worst case scenario."
Artem Bityutskiy replied, "I personally refuse to compare a finished FS with handles all the crucial flash features to a non-finished FS. It just makes no sense. LogFS was talked about back 2005 in Linux Kongress, but is not finished yet. Let's talk about it when it is production ready."
From: Jan Engelhardt <jengelh@...>
Subject: Re: [RFC PATCH] UBIFS - new flash file system
Date: Mar 31, 8:29 am 2008
On Thursday 2008-03-27 15:55, Artem Bityutskiy wrote:
>
> here is a new flash file system developed by Nokia engineers with
> help of the University of Szeged. The new file-system is called
> UBIFS, which stands for UBI file system. UBI is the wear-leveling/
> bad-block handling/volume management layer which is already in
> mainline (see drivers/mtd/ubi).
>[...]
And how does it compare to logfs?
--
From: Jörn <joern@...>
Subject: Re: [RFC PATCH] UBIFS - new flash file system
Date: Mar 31, 9:40 am 2008
On Mon, 31 March 2008 14:29:59 +0200, Jan Engelhardt wrote:
> On Thursday 2008-03-27 15:55, Artem Bityutskiy wrote:
> >
> >here is a new flash file system developed by Nokia engineers with
> >help of the University of Szeged. The new file-system is called
> >UBIFS, which stands for UBI file system. UBI is the wear-leveling/
> >bad-block handling/volume management layer which is already in
> >mainline (see drivers/mtd/ubi).
> >[...]
>
> And how does it compare to logfs?
Both share similar design goals. Biggest difference is that ubifs works
on top of ubi and depends on ubi support, while logfs works on plain mtd
(or block devices) and does everything itself.
Code size difference is huge. Ubi weighs some 11kloc, ubifs some 30,
logfs some 8.
Ubi scales linearly, as it does a large scan at init time. It is still
reasonably fast, as it reads just a few bytes worth of header per block.
Logfs mounts in O(1) but will currently become mindbogglingly slow when
the filesystem nears 100% full and write are purely random. Not that
any other flash filesystem would perform well under these conditions -
it is the known worst case scenario.
Jörn
--
Victory in war is not repetitious.
-- Sun Tzu
--
From: Adrian Hunter <ext-adrian.hunter@...>
Subject: Re: [RFC PATCH] UBIFS - new flash file system
Date: Mar 31, 8:47 am 2008
Jan Engelhardt wrote:
>
> On Thursday 2008-03-27 15:55, Artem Bityutskiy wrote:
>>
>> here is a new flash file system developed by Nokia engineers with
>> help of the University of Szeged. The new file-system is called
>> UBIFS, which stands for UBI file system. UBI is the wear-leveling/
>> bad-block handling/volume management layer which is already in
>> mainline (see drivers/mtd/ubi).
>> [...]
>
> And how does it compare to logfs?
We don't know a lot about logfs, so you will really have to make
your own comparison. However our general impressions are as follows:
1. In our testing logfs file operations seem to be much slower,
see http://osl.sed.hu/wiki/ubifs/index.php/IOzone
2. logfs code base is much smaller i.e. UBIFS has 3-4 times as many
lines of code.
3. logfs does not seem to have bad-block handling.
4. logfs does not seem to have wear-leveling.
5. We are not certain how scalable logfs is.
We could be wrong about those things - don't flame us if we are.
Ask us about UBIFS, not logfs.
--
From: Jörn <joern@...>
Subject: Re: [RFC PATCH] UBIFS - new flash file system
Date: Mar 31, 9:20 am 2008
On Mon, 31 March 2008 15:47:05 +0300, Adrian Hunter wrote:
> >
> >And how does it compare to logfs?
>
> We don't know a lot about logfs, so you will really have to make
> your own comparison. However our general impressions are as follows:
>
> 1. In our testing logfs file operations seem to be much slower,
> see http://osl.sed.hu/wiki/ubifs/index.php/IOzone
Shiny numbers! Performance has improved significantly in the last six
month. Still worth a closer look.
> 3. logfs does not seem to have bad-block handling.
Bad blocks at mkfs time are handled, blocks turning bad later on aren't
yet.
> 4. logfs does not seem to have wear-leveling.
It does.
Jörn
--
Fools ignore complexity. Pragmatists suffer it.
Some can avoid it. Geniuses remove it.
-- Perlis's Programming Proverb #58, SIGPLAN Notices, Sept. 1982
--
From: Artem Bityutskiy <Artem.Bityutskiy@...>
Subject: Re: [RFC PATCH] UBIFS - new flash file system
Date: Mar 31, 10:00 am 2008
Jörn Engel wrote:
>> 3. logfs does not seem to have bad-block handling.
>
> Bad blocks at mkfs time are handled, blocks turning bad later on aren't
> yet.
I personally refuse to compare a finished FS with handles all the
crucial flash features to a non-finished FS. It just makes no sense.
LogFS was talked about back 2005 in Linux Kongress [1], but is not
finished yet. Let's talk about it when it is production ready.
[1]. http://www.linux-kongress.org/2005/abstracts.html#4_4_2
--
Best Regards,
Artem Bityutskiy (Артём Битюцкий)
--
