Re: Compression Performance Results

Hello,

That's interesting, but this type of test is very sensible to the data 
you backup. You could probably get completely different results with 
many small files or few big files, as well as source code in place of 
jpeg pictures...
I'm surprised by the fact that you get a bigger archive when adding -Z 
... option, as well as it takes less time for bzip2 to compute than for 
gzip.... strange. Maybe the fact you probably have many small files made 
the difference... compression buffers must be flushed at each file 
boundary to be able to restore files one by one if necessary. This has a 
cost: a less good compression ratio. I also noticed that it is useless 
to try to compress a file smaller than few hundred bytes, thus if you 
have many of theses small files, probably bzip2 behaves less well than 
gzip, which is obviously not the case for large files.

Regards,
Denis.

Tristan Rhodes wrote:
> Greetings,
> 
> Here are my rather surprising results of different compression settings in Dar.  I always rebooted before starting each test.  My hardware is a Pentium III 533 MHz, 640 MB RAM, and two 7200 RPM IDE hard drives.  Each test was a complete backup of a basic Suse 8.2 linux install.  I made the backups onto the second hard drive.
> 
> No compression
> dar -c /mnt/backup/nozip -R / -P var/ -P proc/ -P mnt/backup -D
> Time:	10 minutes
> Size:	1,499,000,000
> 
> Gzip
> dar -c /mnt/backup/gzip2 -z -R / -P var/ -P proc/ -P mnt/backup/ -D
> Time: 70 minutes
> Size:	614,000,000
> 
> Gzip (Don't compress *.gz or *.bz2)
> dar -c /mnt/backup/gzip-skip -z -R / -P var/ -P proc/ -P mnt/backup/ -D -Z *.gz -Z *.bz2
> Time:	72 minutes
> Size:	617,000,000
> 
> Bzip2
> dar -c /mnt/backup/bzip2 -y -R / -P var/ -P proc/ -P mnt/backup/ -D
> Time:	44 minutes
> Size:   591,000,000
> 
> Bzip2 (Don't compress *.gz or *.bz2)
> dar -c /mnt/backup/bzip2-skip -y -R / -P var/ -P proc/ -P mnt/backup/ -D -Z *.gz -Z *.bz2
> Time:	40 minutes
> Size:	589,000,000
> 
> These results show that Bzip2 (skip compressed files) appears to be the better compression choice.  I have not tested perfomance during a restore.  
> 
> Please comment on these results, and compare with your experiences.  Theoretically, Bzip2 should have been much slower, with a small advantage in compressed size.  
> 
> Tristan Rhodes
> 
> 
> -------------------------------------------------------
> This SF.net email is sponsored by OSDN developer relations
> Here's your chance to show off your extensive product knowledge
> We want to know what you know. Tell us and you have a chance to win $100
> http://www.zoomerang.com/survey.zgi?HRPT1X3RYQNC5V4MLNSV3E54
> _______________________________________________
> Dar-discussions mailing list
> Dar-discussions-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
> https://lists.sourceforge.net/lists/listinfo/dar-discussions
> 
> 
> 




-------------------------------------------------------
This SF.net email is sponsored by OSDN developer relations
Here's your chance to show off your extensive product knowledge
We want to know what you know. Tell us and you have a chance to win $100
http://www.zoomerang.com/survey.zgi?HRPT1X3RYQNC5V4MLNSV3E54