tar compression file size(s)

[MikeFinch]

Hi All!
I have just used Dolphin to compress a folder. I right clicked the folder and chose 'Compress/Compress To...' and saved the .tar.gz file to the Home (same) directory. The size of the tar.gz file is 127.8kb.
I then saved the IDENTICAL folder to my external hard drive and the tar.gz produced is only 85.0kb.
Why the whopping difference?
Mike

[nigelc]

Hi,
What sort of data was/is it? It might have had lots of space characters in it, or repeating data.

[doktor5000]

?
How does that matter for the created archive size, when using the same source data, but two differerent source media/filesystems creates different size archives?

[MikeFinch]

Hi Nigelc!

The files were/are php scripts. Yes lots of white space I suppose. Is it a coincidence that the largest was almost exactly 50% bigger than the smallest?

Mike

[nigelc]

Mike,
It makes quite a difference whether the files are random data.
These two files are the same random data. One is compressed and one not. The compressed is really bigger than the original.

Code: Select all

-rw-r--r-- 1 ewok ewok 1073915443 Nov 21 15:19 file01.gz
-rw-r--r-- 1 ewok ewok 1073741824 Nov 21 15:16 file-xx


This the same size file, but all zeros.

Code: Select all

-rw-r--r-- 1 ewok ewok    1042077 Nov 21 15:23 zero-xx.gz


[doktor5000]

Well, that's logical but not what's described in the initial post, or at least not how I've understood it.

OP says he compressed one folder to two different filesystems, getting different sizes of the generated tarball.
Differing factor is the filesystem, not the data from which the tarball is created which is identical as mentioned by OP.

[MikeFinch]

Folks! Yes the files were identical. Still a 3:2 ratio in sizes and I still don't understand why.
Ah well!
Cheers.

[doktor5000]

If you don't provide informations about the underlying filesystems of the the folders where you copied the tarball to, nobody can tell the cause of the differing size.

[MikeFinch]

doktor5000

I think the file system in both cases was ext3.

Cheers!

[doktor5000]

Could you show the output of

Code: Select all

tune2fs -l /dev/sdxN

for both filesystems?

[nigelc]

How would the type of filesystem affect the the compression ratio?

[doktor5000]

Differing inode sizes, block size and bytes-per-inode ratios ...
And it's not the compression ratio that's affected, but simply the resulting filesize of the tarball.

[MikeFinch]

Doktor5000

Haven't had a go using tune2fs etc but was wrong about the HD filesystem.

My Desktop PC is ext3 and the external HD is NTFS. Maybe that is the answer? Therefore the smaller one was NTFS.

Does this mean ext3 is less efficient (in terms of filesize)? Hope not.

MikeFinch

[doktor5000]

So the differing factor is in the underlying filesystems. And that does not mean that NTFS is more efficient,
that only means the default filesystem configuration in your case provides better results with small files with NTFS with your workload.

Please mark the thread accordingly by editing the topic of the first post and prefix it by [SOLVED], thanks