remote site client deduplication with fingerprint seeding

[FlyMountain]

We are on NBU 7.6 and planning to implement remote sites client deduplication.  We don't have any media servers on remote site. I have a few questions about seeding fingerprint. thanks in advance.

1). Can we simply copy data from all remote clients to an USB device then ship the USB to our data center for backup and seeding?  Does the data structure (file/folder pathes) matter for seeding?

2). I am assuming all backup images which will be used for fingerprint seeding, should be either client dedu based backup or media server dedu based rather than normal backup. Please confirm.

3). If we have both client or storage server seeding avaiable, which one is better?  Since each remote site may have multiple clients, i am thinking to use storage server seeding. Please confirm.

[GulzarShaikhAUS]

Are you using Appliance? If yes please give the details and if no then please give the details about MSDP server

[Andrew_Madsen · Accepted solution]

To answer your questions in order:

1) Yes you can copy the files to a USB, ship that to your central site, attach it to an appropriate client, back up the folders in a special backup job, and then point the remote client to that image. You would use the pd.conf method outlined in this technote: https://support.symantec.com/en_US/article.TECH144437.html This technote has both option you are asking about. however they have their individual issues.

2) This would only work if you are backing up to a deduplication pool using the client side deduplication.

3) As I said before both processes have their strengths and weaknesses. The pd.conf method calls for either a copy of all the machines being sent to the central location or hoping that a few of the images would be enough to help with deduplication. If you use the seedutil tool you once again need to find an image that would have the best chance to deduplicate but is faster to configure for multiple clients.

Whichever method you chose read the technote carefully because there are different variations to the methods depending on the version of NBU you are running. Also what is missed is the need to turn on client side deduplication through Host Properties of  the master server. It is off by default.

[FlyMountain]

Thanks Andrew. You did provide very useful information.

One more question, when we copy the data from differnt clients to USB device on remote site, do you have to keep all data structures (paths, timestamps...) or dedu agent is smart enough to find reference files from seeding fingerprint? thanks again.

[FlyMountain]

No appliance. Regular puredisk MSDP on SAN disks. I was thinking to create different MSDP for different remote sites on same MS, but looks like only one DP is allowed on same storage server. thanks.

[Andrew_Madsen]

You do not need to keep the data paths. You can attach the USB which becomes a new drive letter on the client you are going to use as the backup client. That way if all your data does not fit on a single USB you can put the remainder on a second USB and when you attach it to the target client you set the policy to backup drive 1 and drive 2.