How does NDMP work?

http://lmgtfy.com/?q=How+does+NDMP+work%3F

I love lmgtfy.com LOL.

That said, NDMP is a protocol to control typically NAS devices.  It's not to control a data format, or control how the data is extracted from the device, think of it as SCSI commands sent via ethernet to a NAS device to do something.  Each NAS vendor implements the NDMP protocol differently.  EMC used to TAR their data, Isilon does something different, as does BlueArc.  One NDMP backup may not be possible to be restored to different hardware, so in a way look at NDMP at proprietary.

NetApp does a UFS_Dump I believe of the data, and sends it to your target, which in this case is over the LAN to your media server, to direct attach tape.  In some configs, you can allow your NAS to write directly to tape with the tape drives on a SAN along with your NAS, or your NAS has some SCSI ports on the back of it to accept a SCSI attached tape.  NBU would tell the NAS via NDMP to move LUN 1 to SCSI ID X, and the device would report back in NDMP what it did, and other status info.

In your case, since the data is deduped by NetApp/ONTap, My guess, is that the data is being re-hydrated to it's original size as it's being sent over the wire to your target.  

LOL - brilliant - that lmgtfy.com gag was classic. Nice one Zak... hold on I've got to go and dry my eyes..........I'm back. I bet the long winter nights fly by in your house.

Zak, no more funnies like the previous one or my ribs are in danger of cracking.

Thanks for the reply teiva-boy. I probably wasn't that clear with my question. My question was based around the capacities I'm seeing on my volumes and the difference between them and what is actually written to tape.

"What is the 2200GB written to tape made up of?" - The 2200GB is what the NBU Actvity Monitor says has been writtent to tape. The used space within the volume is less than this, i.e 1640GB? So why is the data written to tape more than the used space from within the volumes.

Thanks.

Hi,

I'm not 100% sure of this but I suspect that the NDMP backup works the same way as a RAW backup. In NetBackup you can configure a backup policy to backup the RAW data i.e. the file system or blocks rather than the files residing on the file system. This is handy when dealing with file servers with a lot files. Backing up the blocks gives you a huge performance increase.

The problem is with this type of backup the whole file system is backed up each time, so if its 100GB, and you've got 20GB worth of files, the backup would be 100GB each time.

I think this is because from the netapp side each CIFS share or volume that you create is actually a file system by itself, you're backing up the whole file system RAW, and getting the increased backup size.

You might want to wait for a NDMP expert to comment :)