Data Domain with Storage node

It will back up the same amount of data as the deduplication takes place on the storage node and not on the client - the amount of data sent from the client to the storage node will thus be the same for a full backup...  Deduplication will take place on the storage node with DDBoost enabled, however it does not necessarily follow that if you're data changes say 10% it will only take 10% of the time as before as there is an overhead in the processing of the data, however you should see a performance increase as the processing overhead for deduplication is less than the overhead of putting all the data on the wire (at least that is the theory!).  If you could give some info on your environment and the type of data/amount of change we are seeing it may be easier to comment further...

Hi David,

Thanks for the reply. The concern here is that, it’s a dedicated storage node, there is no client configured through this storage node. The problem is, I took a full backup of the storage node last Saturday and it completed successfully. But when again started the full backup of the storage node on Monday, it took approximately the same amount of data and time. As you updated the storage node should perform a de duplication and send only the changed data. But it’s not happening.

Hope my explanation was proper.

The backup itself will actually take the same amount of time from what I've seen, where your "savings" comes in with data domain is that it won't be sending as much information from your dsn to the data domain.  Networker has reporting built in that will show how much your dedupe ratio should be for a particular backup job.  You can also look at that info from the data domain side, but it won't give you the dedupe of a certain backup that ran, it will give you the dedupe of everything sent to it.

From the way it was explained to me you should see dedupe even during your first full backup.

I know this isn't comparing apples to apples, but it gives an idea of how the data domain dedupe works:

If you were backing up the sentence: "This is my first backup", this is the "unique" data that should be sent to the data domain:

"This myfrbackup

I know data domain deals with a lower level than that, but you should be able to get the idea from that.  It's already backed up the letter "i" so it won't backup the letter "i" that's in the word "is" again.

Hi Vishal/David

Have similiar setup and backup doesn't happen any quicker. Same all the time on all apps. However, reduction can be seen on the amount of size stored on DD. Was curious to find out about the backup time too..however, hv self read/taught that the logic is that all data actually sent to DSN-DDBoost to filter/sort out(processing center-backup time)...which means, it still has to do the treewalk from data received and compare what changed and not....the changed data then would be sent to DD which is way lesser amt of data than sent to DSN.

If anyone sees faster backup, please do share... thanks..

If DD boost happens on storage node, I assume data is still being sent to storage node and only processed there. With that in mind, there are instances where Boost is processed on client side - for example certain newer modules which support it. As for file system backup, as they say, watch this space...

All of the data is still sent to the storage node and only the new deltas are sent to the dd box.  However you can make the file system server a dedicated storage node and only new data will be sent to the dd box.

When using a DataDomain with DDBOOST the segmenting and fingerprinting is done on the storage node. The storage node will only send these blocks to the DD which are new for the DD. So the deduplication process is not done only at the DD - it is distributed to the storage nodes.