[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [rdiff-backup-users] Restarting development ... or starting over

From: Nicolas Jungers
Subject: Re: [rdiff-backup-users] Restarting development ... or starting over
Date: Tue, 06 Apr 2010 15:01:07 +0200
User-agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv: Gecko/20100404 Lightning/1.0pre Thunderbird/3.0.4

On 04/06/2010 02:34 PM, Daniel Miller wrote:
I wasn't really prepared to make this announcement so soon, but
now seems like a good time to let the community know. I've been
working on a new implementation of rdiff-backup since about a
month ago when


I'm not a coder... [snip]

Now, on top of that I'd like to have all the fanciness present in
all the other programs, particularly a SQL backend for the metadata

Out of curiosity, if you're not a coder what do you plan to do with a
SQL backend?

I used Bacula for a while and the possibilities of the SQL backend seduced me.

Anyway, decoupling the content and all the metadatas including storage path, slicing, encryption, checksum really appeals to me. So it's not really the SQL backend that I want, but the ability to query the repository with the needed criterion, even to simply shows management what the backup systems does.

Now my use of rdiff-backup is in no way limited by space but by bandwidth, I do two remote copies of the repository in addition to the local one.

I actually considered using SQLite as a backend for storing metadata,
but I just couldn't justify it since rdiff-backup would have no real
benefit from the features of SQL (complex queries, joins, etc.). The
only real benefit would be simpler random-access (for restoring
selected files, for example). However, random access in a database
either requires an index, which takes extra HDD space, or it involves
a sequential scan, which is exactly what rdiff-backup already does.
Why add another layer of complexity if it's not needed? It would
likely just slow things down and add bloat to the repository. Having
said all that, I did write the new version with a well-defined
"database" interface, which means other backends (possibly MongoDB)
could be added without affecting the frontend.

Out of curiosity, is your code accessible and usable?


reply via email to

[Prev in Thread] Current Thread [Next in Thread]