Re: [Bacula-users] Plans for support block-based dedupe?

On 04.01.2013 12:42, Radosław Korzeniewski wrote:

Hello,

2013/1/4 tonyalbers <bacula-forum AT backupcentral DOT com>

Hi Silver,

The downside of client-side dedupe is actually a combination of several things.

1. The client does the dedupe calculations etc. this means that it takes CPU resources on the client which it might not have.

Yes it could be a problem in some cases. Our testings shows that deduplication computing has less cpu requirements than compression or encryption. Using a modern processors available today in the market this should not be a problem.

2. Sometimes, client-side dedupe is chosen because of bandwidth limitations to the backup server (remote offices etc). In case of a large restore all files will have to be sent from the backup server to the client anyway, and this can be almost impossible because of the time it will take. That's the RTO side of it.

The same for server side only deduplication or no deduplication at all. There are no difference. If you have a slow link that your recovery will be slow.

This is also the case only with a complete disaster recovery (ie. not a partial restore) and this is no better without client-side deduplication - the only thing is that in that case you have to really minimize your backup data (which you could also do with client-side deduplication, but you wouldn't be forced to do that).

3. Partial restores can only be made to the client that did the backup, not to another client if necessary(redirected recover).

Not true. In our implementation we can restore any file to any deduplication enabled bacula client. No restrictions here.

He meant PARTIAL restore and in this case it's totally valid. But I don't think it can be considered as a downside of client-side deduplication as such - partial restore is always possible only to the client that already has some portion of the original data.

--
Silver

There are probably more, but these are just what I can think of straight away.

The only real problem could be a #1 above. If your client does not have a sufficient computing power then deduplication could be slow.

The main advantage of client side global deduplication is low network bandwidth requirement with the same deduplication ratio (or better depending on architecture and hardware) as with server side deduplication only which leads to short backup window.

------------------------------------------------------------------------------
Master HTML5, CSS3, ASP.NET, MVC, AJAX, Knockout.js, Web API and
much more. Get web development skills now with LearnDevNow -
350+ hours of step-by-step video tutorials by Microsoft MVPs and experts.
SALE $99.99 this month only -- learn more at:
http://p.sf.net/sfu/learnmore_122812

_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users