performance of rsync

jw schultz jw at pegasys.ws
Fri Aug 8 06:52:09 EST 2003


On Thu, Aug 07, 2003 at 03:06:46PM -0500, Carniado Rodríguez Claudia wrote:
[reformatted due to excessive line length]
> I have a question about rsync, which is new for me. Hope
> you can help me:
> 
> I have copied a binary file whose size is about 100 MB
> using rsync and I noticed that it takes a lot of time.
> The same size with a text file wasted really much less
> time.  That binary file is incremental, i.e. it started
> with a size of 2 kb and then it grew.
> 
> But now I am confused with the following:
> 
> Is the performance of rsync better with no-incremental
> files?

Not at all.  Rsync performs best with incremental files.

> Is the performance of rsync better with text files?

What is a text file?  What we think of as text, rsync treats
as binary.  There is no difference.

> Is it the best tool for binary-incremental files? And if
> it is, which parameters do I have to use on rsync?

As far as i can tell (i'm not omniscient) rsync is the best
tool available for modified (incremental or otherwise)
files.  If you read the documentation and this list you will
quickly find that the primary reason for rsync is modified
files.

You don't say anything about how you are using rsync or the
nature of the file changes.  There is often a big difference
between how a text file is modified and a how a binary one
is.  If the modification of the binary file alters the
entire file rsync won't be able to accelerate it.  The same
can be said of "text" files but few text file modifications
change the entire file.  Try inverting case or rot13 on the
text file and see what effect that has on speed.

Your topology can make a big difference in the effectiveness
of the rsync algorithm.  By default rsync uses the rsync
algorithm to increase the disk load in order to
lighten the network load.  If your disk subsystem isn't
quite a bit faster than your network this will actually be
slower.


-- 
________________________________________________________________
	J.W. Schultz            Pegasystems Technologies
	email address:		jw at pegasys.ws

		Remember Cernan and Schmitt



More information about the rsync mailing list