Pietro Battiston

Questa è una vecchia versione del documento!

I still haven't found the perfect backup solution for a Linux desktop. This page reports my findings.

But first, a list of my requirements, since yours may differ:

graphical interface, since my non-hacker relatives must be able to use it easily ¹⁾ preferably gtk-based
ability to backup through smb
ability to browse backups and restore single files

I admit it: I'm just looking for something like Apple Time Machine, OK?

Here are the evaluations I made (in green and red, positive and negative aspects that I consider particularly relevant in the comparison):

	backintime	flyback	deja-dup	simple backup	timevault	fwbackup
GUI	nice	<html><span style=“background:#00FF00;”>very nice</span></html>	<html><span style=“background:#00FF00;”>very nice</span></html>	<html><span style=“background:#FF0000;”>no GUI while backupping (so no errors reported!). All preferences are lost if the user clicks “use recommended backup settings”!</span></html>	?	nice (judging from screenshots)
shows progress during backup	only file currently copied	only time elapsed (and that may be an intrinsec limitation of being based on git)	?	no (how could it without GUI?)	?	?
exclusions handling	takes apparently any regular expression - though the GUI is not very clear on that	<html><span style=“background:#FF0000;”>few preset exclude options (impossible to exclude a single subfolder)</span></html>	?	<html><span style=“background:#00FF00;”>free choice of inclusions and exclusions</span></html>	?	only provides a list (presumably of single folders/files)
can browse backupped files	?	yes, but only a list of files is provided (not organized in folders)!	<html><span style=“background:#FF0000;”>no!!!</span></html>	yes, but each incremental backup only shows files backupped at that time	?	?
can restore a single file	?	<html><span style=“background:#00FF00;”>yes</span></html>	<html><span style=“background:#FF0000;”>no!!! the only available operation is “restore the whole backup”!</span></html>	<html><span style=“background:#00FF00;”>yes (though the operation is not very fast if the file is part of a large backup)</span></html>	?	?
backup targets	only filesystem	only external filesystems	filesystem, smb?	filesystem, ssh, ftp	?	filesystem, ssh, maybe others
speed	?	<html><span style=“background:#FF0000;”>very slow</span></html>	good	very good	?	probably depends on “engine” used
space occupied	<html><span style=“background:#FF0000;”>much (no compression)?</span></html>	<html><span style=“background:#00FF00;”>optimal (no replication, and only updates of non-binary files are recorded)</span></html>	?	good (no replication, gzipped)	?	probably depends on “engine” used, but maybe there is no attention to replication?
other pros		<html><span style=“background:#00FF00;”>can show differences between versions of (non-binary) files</span></html>	<html><span style=“background:#00FF00;”>nice notifications mechanism</span></html>		?
other cons		<html><span style=“background:#FF0000;”>Doesn't keep track of emtpy folders and folders/files permissions, doesn't allow deleting old backups. Not available in Debian repositories, neither as a tarball! Available only as a (low quality) .deb</span></html>		<html><span style=“background:#FF0000;”>(at least by default), only runnable by root</span></html>	<html><span style=“background:#FF0000;”>doesn't work with python 2.6! Last commit (as of April 2010) was in 2007!</span></html>

Other apps I may want to look at: zink, grsync, areca.

Contributions and suggestions will be appreciated: toobaz [AT] email [GUESS WHAT… DOT!] it

¹⁾

Though I've been pointed at cron-based solutions as the ones providing less hassle for unexpert users, I do not share this view since unexperted users may be perfectly able to plug in a USB hard disk and pretend that at that time a backup occurs, and they can unplug the USB hard disk at the end.