Deduplication Ratio Analysis

[richreitenauer]

I am trying to interpret my deduplication ratio.  I have a EMC Data Domain 670 device and running a backup with 3 Server 2008 R2 test machines.  Currently I am getting a deduplication 9.26:1.  My data domain shows pre-comp and post-comp, can that be interpretated as deduplication?

Thanks.

[CraigV · Accepted solution]

The more you run your dedupe job, the higher the dedupe ratio.

I'd suggest taking a look at the document on the Symantec site below which shows you how to get the most out of dedupe. Check the link below:

http://www.symantec.com/business/support/index?page=content&id=TECH142668

You might want to check out the MS hotfixes referenced on the Symantec TN below:

http://www.symantec.com/business/support/index?page=content&id=HOWTO21767

Thanks!

[teiva-boy]

Correct from CraigV.  The more you run your backups, and more retained, the greater commonality seen by the DD670 unit.  Thus higher dedupe rates.  Do 10 backups, one after a another, you'll see a better rate than 3 backups.

[richreitenauer]

I wasn't going to use the Deduplication storage folder, and was looking at the client side dedupe though I wanted to backup multiple servers at once.  I have been adding more servers into my backup and the deduplication ratio went down to 6.17:1.  I noticed last night with the three servers that the dedupe ratio went up to 20.81:1.  I am going to add the Sharepoint and AD options to also do media server deduplication.  

Thanks for the documentation.

[teiva-boy]

Wait you are using Symantec Client dedupe, and saving it to a datadomain?  Or are you using datadomain as a backup target?

I can pretty much assure you the DD dedupe is worlds better, faster and more reliable than the BackupExec dedupe.

[richreitenauer]

I am not using Symantec to do the dedup, I have the data going straight to the DataDomain (Set as backup Target).  I guess that Symantec will still do dedup on Exchange and AD since the backup job is set to dedupe on Media Server then send to the Data Domain.  Oh yeah, for the money, the DD is worth every penny.

[CraigV]

...can't wait for our DataDomains to come in early next year...they will solve a lot of our issues with large DBs being backed up!