Questions about de-duplication

[robnicholson]

We've just upgraded to 2010 and renewed our support agreement and I was interested in looking at de-duplication. I was a bit disappointed when this was an additional feature for which we don't have a license - is that correct?

If so and before purchasing the license, I have a few questions about de-duplication:

1. We have a 6TB SATA backup-to-disk system. We carry out a full backup every week and a differential every day. We get about six weeks back of backups. Am I right in saying that with de-duplication, we'd make more effective use of this backup space? For example, assume the same 2GB video is there each week. Without de-duplication, I'm assuming that ~6 copies of this are stored in the backup. WIth de-duplication, one one copy would be stored in the B2D system? This is an attraction as it would increase the capacity/window without upgrading the hard disks
2. Is the de-duplication reduction on an existing B2D system pretty much instant or does it take a while to ripple through. I guess I'm asking is does some program run at the start to sort out existing duplicated files or does it rely upon the old duplicated files to ripple through and fall out of the B2D window?
3. I'm at a bit of a loss to see how de-duplication would work with backups to tapes. With B2D, all the backup media is online and available so the system knows whether the original copy of the file is there. With tape, what happens if the tape holding the only copy goes bad or missing? Considering how flaky our Quantum SuperLoader 3 has been, I'd be very worried about there been only one copy in the backup system. Also, it would make a full restore of the last backup tricky. At the moment, we just bring in the tapes since the last full backup. With de-duplication, it could suddenly ask for a very old tape in any weird order as that's where the copy as stored. Am I missing the point here?

Cheers, Rob.

[nhej-]

check this links maybe it can help
http://seer.entsupport.symantec.com/docs/340834.htm
http://seer.entsupport.symantec.com/docs/342718.htm
http://seer.entsupport.symantec.com/docs/338254.htm
http://seer.entsupport.symantec.com/docs/342709.htm

[teiva-boy · Accepted solution]

1.  Yes, pretty much as you described.  

2.  A full backup must happen, you will get little to no dedupe on the first pass, but all subsequent passes you'll see a savings in storage consumed when it goes to the Deduplication storage folder, which is a different target than your current B2D folder.  

3.  There is no Dedupe to tape from Symantec, and wont be for a while.  For recovery reasons as you've described, it was not implemented.  

[FischerRoman]

Keep in mind that DeDuplication is some really nice feature, but still has some bad stability problems.
I still have a case open since about a month with some bad corruptions, not-being-able-to-restore-from-dedup-store, broken communication between dedup-device and backup exec media server, etc.

I would REALLY recommend NOT to use dedup for productive environments until BE2010 R2 will be released (which could happen somewhere in september or october 2010).

But make yourself a picture, read about some of the dedup issues at the forums.
Also take a look at the recently released hotfixes and fixed issues here: http://www.symantec.com/business/support/downloads.jsp?pid=15047&type=patch


You could also install the DeDuplication feature and try it out for 60 days without any license - legally!
Maybe you're lucky and it works as it should - and stable.
If not, you didn't spend money for a not-so-well-working feature.

If it works perfectly, go ahead and buy a license for DeDuplication! ;)

[robnicholson]

Thanks everyone for the comments. I've recently installed a trial version of BE 2010 to have a play myself and I now have a much greater understanding of how it works.

In order to gain a deeper understanding, I'm going to stick some large TB hard disks in my PC and leave the virtual trial media server backing up our real data.

I was surprised to learn on ServerFault of the RAM requirements for deduplication. Glad I read that before sticking it on the production system!

Cheers, Rob.

[robnicholson]

>I would REALLY recommend NOT to use dedup for productive environments until BE2010 R2 will be released (which could happen somewhere in september or october 2010).

The version I'm using for trial is BE 2010 R2.

Cheers, Rob.