Backup Exec 2010 Deduplication Set Up and performance

[JGooderh69]

Hello All;

I would like some feedback on the following scenario. We are undergoing a refersh of our Novell Netware environment and it's dovetailing into a new implementation of Backup Exec 2010.

We have implemented SLES/OES with NSS volumes. I have installed the Linux agent on the servers. There is a new install of Backup Exec 2010 installed on Windows 2008 Standard 64bit. I have a separate partiton of a RAID 5 violume designated as deduplication folder. (it's a 800GB raid 5 with 100Gb dedicated for C: and 700GB dedicated for D:). We have purchased the DeDuplication option to back up two remote servers over WAN links.Iknow that the dedup folder should be a separate volume, does it need to be physically separate?

Do I define the jobs the same as I would backing up to tape (full Fri. nite and incrementals M-R)? As it is right now, they are. The back up everything on the server (OS and NSS volumes, not the iFolder storage however). When the first incremental ran, it ran pretty much as a full and I expected that since it was all new data to the server. The full ran and backed it all up again, then the incremental is also backup it all up over again. What have I configured wrong? Also, performance over the WAN links degrades over time. It will slow to 23MB/min.

I've tried looking for some docs online and I have found the best practices docs. They are not granular enough to define the job, retention level, etc..

Let me know your thoughts.

Thanks.

[CraigV · Accepted solution]

Hi,

If you could run it on a physically separate HDD, that would be better. At least if you lost the drive your OS ran on due to a hardware failure, you'd have your dedupe folder sitting elsewhere. If you're unable to, I suppose you're left with running it on another partition on the same drive. Just bear in mind you can't move the dedupe folder to another drive/folder AFAIK...

Setting up a backup job to run to a dedupe volume, a B2D or tape is exactly the same between them. You define where the job runs to during the wizard itself. No difference there, and no special configurations during the initial setup.

As far as how dedupe backups run, the first is full, and subsequent jobs are incremental. From my testing (we don't use Dedupe...don't ask me why!), the incremental jobs all appear to be FULL backups. However, if you go into the job, or check the job stats on the Job MOnitor page after it finishes, you'll see what deduplication ratio you will getting. If you've only got 1:1, then there is a problem. But the more time it runs, the better the dedupe ratio, and the less data you're effectively backing up each night.

the BP documents will give you what you should be doing...so for example, running on an x64 system, maybe tuning disks etc. It doesn't give you job retentions, as there is no BP around this. This is defined by you, dependent on how long you need to keep data for, as defined by your company!

Thanks!