Just Launched: You can now import projects and releases from Google Code onto SourceForge
We are excited to release new functionality to enable a 1-click import from Google Code onto the Allura platform on SourceForge. You can import tickets, wikis, source, releases, and more with a few simple steps. Read More
From: Vahe Chahinian <vahe.chahinian@co...> - 2013-05-09 19:20:40
I am trying to find out if I should use a Bacula client (Fd) on our
fileserver or just mount the folders we want to back up onto our bacula
server and then backup those folders locally. I assumed that the hashing
would be done on either the client or the server. Ideally I would like
to put the least amount of strain on the file server. Also does the
client hash the file to be backed up and send it to the server to see if
the data has to be sent? If that is the case then it might make more
sense to just do the hashing on the file server to save a lot of network
2013/5/9 Vahe Chahinian <vahe.chahinian@...>
> I am trying to find out if I should use a Bacula client (Fd) on our
> fileserver or just mount the folders we want to back up onto our bacula
> server and then backup those folders locally.
Both cases are good and will work. Both have advantages and disadvantages,
i.e. NFS is not as efficient as Bacula communication protocol, especially
when you will backup a lot of files (a fileserver, in example :) ), but if
you need to compress or encrypt your data then using NFS allow you to move
compress or encryption computations from fileserver to backup server, etc.
You have to check what is best for you.
> I assumed that the hashing
> would be done on either the client or the server.
What hashing do you mean? Currently Bacula can perform MD5 or SHA1 file
digest computation during backup. It is performed on client side and cannot
be moved to backup server.
> Ideally I would like
> to put the least amount of strain on the file server.
Great, then NFS will be your friend.
> Also does the
> client hash the file to be backed up and send it to the server to see if
> the data has to be sent?
It sounds like block level data deduplication which is not currently
> If that is the case then it might make more
> sense to just do the hashing on the file server to save a lot of network
Yes it is. It is called "deduplication on source". Bacula currently has
deduplication on source (base jobs) but it works completely different then
BTW. You should ask above questions which are not related to Bacula
development on bacula-users list instead of bacula-devel. Do this next time.