0

As title, I wonder how fsutil in windows can create a really large file so fast. Does it really allocate real cluster for that file or it just writes down file's metadata? Consider two commands below:

fsutil file createnew testFile <1Tb>
dd if=/dev/zero of=testFile bs=1024M count=1024

So I create a file with 1Tb size, the problem is with fsutil, the file is nearly created immediately, but with dd, it took over 1 hour to complete. Therefore, I guess that fsutil only writes metadata to the file header, the real file cluster will expand whenever needed. Do I think right?

Hoang Thong
  • 95
  • 1
  • 10

1 Answers1

0

from Microsoft documentation

createnew Creates a file of the specified name and size, with content that consists of zeroes.

from here, you can say that the file must be all zeros ([...]with content that consists of zeroes)

but, if this is truth

[...]the file is nearly created immediately[...]

I think that you are right: probably, fsutil creates the file with bytes marked as free at the time of execution, but doesn't write those bytes

When you use dd like this 

dd if=/dev/zero of=testFile bs=1024M count=1024

you are actually writing, "byte by byte", zeros in each byte of the new file

You can do this:

fsutil file createnew testFile_fsutil <1Tb>   #(on Windows)

dd if=/dev/zero of=testFile_dd bs=1024M count=1024   #(on Linux)

and then, you can see the contents of testFile_fsutil on any hexeditor, and looking for non-zero bytes, or, more precisely, from Linux you can do ( 1099511627776 bytes = 1 Tebibyte ):

cmp --bytes=1099511627776 testFile_fsutil testFile_dd

or

cmp --bytes=1099511627776 testFile_fsutil /dev/zero

or even using hashes:

dd if=/dev/zero bs=1024M count=1024 | sha1sum

return

fea475204f4ad1348f21fad930a7c721a2858971

so, 

dd if=testFile_fsutil bs=1024M count=1024 | sha1sum

must return the same.

Note: to prove your point much more quickly, you can use a much smaller file for test.

alsjflalasjf.dev
  • 1,589
  • 13
  • 11