[tahoe-dev] Potential use for personal backup

[Shawn Willden]

On Tue, May 22, 2012 at 9:52 AM, Saint Germain <saintger at gmail.com> wrote:

> On 22 May 2012 17:44, Shawn Willden <shawn at willden.org> wrote:
> >> If I have my computer at home and a remote server somewhere, I want to
> >> use 1-out-of-2 configuration in case I lost either of them.
> >
> > If you can maintain pretty high uptime and contribute at least 500 GB,
> you
> > might consider joining VolunteerGrid2.  Then instead of 1 of 2 you can
> do,
> > say, 10 of 20.  Or 15 of 20.
> >
>
> Interesting. However I really don't need this level of redundancy,
> pictures of me are not so critical ;-)
>

:-)

A less-obvious implication is that 1-of-2 requires doubling the size of
your files to store them, while 15-of-20 increases their size by only 33%.
 If we had 40 nodes, you could do 35-of-40 and "waste" only 14%.  While
storage is cheap enough that none of that matters, the impact on bandwidth
is quantitatively the same and qualitatively more important.

>> So tahoe-LAFS works in that case basically as Dropbox synchronisation
> >> (with the encryption).
> >> Is there some kind of optimization which only transfer modified parts
> >> to the other nodes, or is the whole file transfer each time it is
> >> modified ? Especially for VM.
> >
> > It's whole-file.
> >
>
> Ok, definitely a problem for VM then.
>

Probably.  I can think of some workarounds, such as snapshotting your VMs,
which causes changes to be written to a small differences file leaving the
bulk of the virtual storage unmodified.  But, no, Tahoe isn't particularly
well suited to backing up large files.


> >> Same question for moved files.
> >
> > Moved files are not re-uploaded.  As long as the content is the same,
> Tahoe
> > will recognize that the file is already in the grid.
>
> Ok good for movies, pictures and mp3 then.
>

Yes.


> >> If I have a VM of 1 Go that I want to regularly backup on the remote
> >> server. Do I get to use 1 Go on the home computer and 1 Go on the
> >> remote computer for the current version and 1 Go for each backup of
> >> the VM ?
> >
> > I don't understand what you mean by "1 Go".
> >
>
> Giga-octets. Sorry 1 Gigabyte then.


Interesting.  That's not an abbreviation I've seen.

Each file you back up is split into N pieces, each of 1/K size, and
distributed to the servers in your grid.  So if you have a 1 GB VM on your
desktop machine and you have two servers in your grid,  and K=1, N=2, Tahoe
will split your VM into two pieces, each 1/1 = 1 x 1 GB in size and store
one in each of your servers, for a total of 2 GB stored to back up the VM.

Or for K=15, N=20, that would be 20 pieces, each 1/15 = .067 * 1 GB = 67 MB
in size, for a total of 1.3 GB stored to back up the VM. :-)

-- 
Shawn
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://tahoe-lafs.org/pipermail/tahoe-dev/attachments/20120522/fdbfd53d/attachment-0001.html>