[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: better install experience


On 11/10/07, Jeff Spaleta  wrote:
> On 10/10/07, Christopher Brown <snecklifter gmail com> wrote:
> >
> > Love to.
> >
> > http://linuxhelp.150m.com/resources/fs-benchmarks.htm

The benchmark on the above page used the ntfs-3g-0.20070118-BETA driver 
(mentioned in the 6th paragraph) and since then tremendous amount of 
performance work went into NTFS-3G. Please see


Nevertheless the driver is still very much __NOT__ optimized but being 
worked on the kernel, FUSE and NTFS-3G level by at least Nick Piggin, 
Miklos Szeredi and the NTFS-3G team.

> > >    Something like:
> > >    http://ntfs-3g.org/performance.html
> > ..which doesn't compare the ntfs kernel driver strangely enough

The kernel driver can't create/delete files hereby it couldn't run the 

However if you look carefully the streaming read/write performances than 
you can notice that that the NTFS-3G performance is very close to the 
performance of the kernel file systems. The reason is simple, typically the 
disk bandwidth is the bottleneck, not the file system quality.

Btw, the above numbers are also outdated and all maintained filesystems 
were improved since then. These numbers are more recent (and not done 
by me ;) though still almost half year old.


> Correct me if I'm reading it wrong... but the information on that 
> benchmark page seems to indicate that the NTFS listings are made using 
> the native Windows NTFS driver running under Windows XP.

Yes, the Windows NTFS driver and NTFS-3G were tested. The kernel NTFS 
driver can't run most of those tests because it can only read files.

> "Interestingly enough, the NTFS/fuse project is most probably another
> effort designed to have NTFS preform poorly under Linux. By design,
> NTFS/fuse is a user space driver, and thus will never compete
> favorably with Microsoft's kernel drivers. 

High performance, block device based hybrid-space file systems are very new 
on Linux and they need quite a lot of optimization work in the kernel, FUSE 
and the file system driver itself. NTFS is also very complex (Windows NTFS 
is about 300,000, ntfs-3g 70,000, ext3 30,000, fat 3,000 source lines 
without comments, etc) and we simply can't do better now.

> The Linux NTFS kernel driver was knocked on the head some years ago, 
> after claims, probably false, of how dangerous it was to use. 

Here the writer confuses Loewis' kernel NTFS driver with Anton's kernel 
NTFS driver.

> Anyway, instead of fixing the supposed problems, the work to that point 
> was just thrown away and the NTFS/fuse project started."

Absolutely untrue. Anton's kernel driver and NTFS-3G have the same source 
base, the later being actively worked on by developers but not the former 
what Anton says he completely rewrote, and it's closed source for now.


NTFS-3G Lead Developer:  http://ntfs-3g.org

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]