[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Mounting NTFS partitions....
On Tue, 1 Sep 1998, Shachar Tal wrote:
> > just as a side note: NTFS is proven slower than FAT.
>
> How did you get to that conclusion? Is there a proof online somewhere?
yes, I don't have the URL handy, but searching through the net will get
you a few sites (try dejanews first).
I talked to Hans Reiser (author of the reiserfs, first for linux and
recently ported to NT, but faught about in court since Microsoft won't
allow him to distribute it or something). Hans made a careful algorythmic
analysis of many file systems, and concluded NTFS is slower than probably
any server filesystem in the market today.
> AFAIK, FAT is based on linked-lists (which are terribly slow - o(n) ),
> while NTFS is a splay tree, with an amortized complexity of o(log n) per
> filesystem access.
becuse of the way it is implemented inMicrosoft code though, it looks like
the OS does twice as many write accesses and realign cycles than linux for
the same amount of data written, don't ask me how (it's not my specialty).
Netware, HPfs and even FAT have been proven faster. you can easely check
it by benchmarking an NT workstation running on the same hardware, once
with NTFS and once with FAT, and see that NTFS is up to 20% slower. my
company never uses it (not because of me, they insisted on NT machines to
be supplied with FAT even before I got here :-)
ok, enough off topic, back to linux (answers and flames in private Email
please)