Deduplication benefits

[nthnu]

We've been using deduplication for a while now. We backup to a NAS, then to an external USB drive. It appears it is helping us save some space for backups. It appears to also increase the time going from NAS to USB due to rehydrating. Right now, our biggest issue is getting things copied quickly. I wanted to know, if we moved away from deduplication and went to a 1:1 backup, would the time from backup to USB improve?

[CraigV · Accepted solution]

Hi,

Maybe, and maybe not. Data would no longer rehydrate, and if you use larger files for B2D you might see a speed increase, but this is at the cost of putting all your eggs in 1 basket. If a larger B2D file corrupts you lose more data.

You would most likely also be unable to keep as many backup sets as you can with dedupe due to the way in which it stores data.

Only way to see is to run a test as no 2 environments are the same.

Thanks!

[jurgen_barbieur]

what are the hardware specs of your server?

what type of USB are you using? (USB 1, 2 or 3)

* if your server's hardware is too small to manage the deduplication and rehydratation, a native backup to disk, duplicate to USB would certainly improve.

* If your server's hardware is ok, I suspect the USB speed to be the bottleneck, and no improvement will be made by moving away from deduplication.

[nthnu]

Ok. Hardware-wise the server is fine. The bottleneck is definitely our 3TB USB 2 drives.

I'm trying to see what can be done to improve backups without a budget. lol

I wouldn't mind a test setup, but we don't have that ability.

[jurgen_barbieur]

is it possible to explain the reason why your secondary copy needs to move faster?

normally this speed is minor compared to the backup itself because this is a secondary copy.

an option could be to afford more usb-drives, attach two or more drives simultanously, and duplicate more jobs simultanously to the different USB drives attached. The load will be spread accross the different usb-drives

[nthnu]

The secondary copy (each server is its own job) currently takes at least three days to copy 2.5TB of data. The next problem resulting from that is if there are errors where we will need to rerun the secondary copy. With only a couple of days left to get it copied, it puts things in crunch time.

I've tried the dual drive setup, but for some unknown reason, I can't get SEPM to do multiple backups to the same drive or multiple drives.