Video Screencast Help
Symantec to Separate Into Two Focused, Industry-Leading Technology Companies. Learn more.

Backup Exec 11D and IBM N3700

Created: 24 Nov 2009 • Updated: 27 May 2010 | 7 comments

Dear all, anyone familiar with IBM N-series storage boxes??
A customer of mine is backing up about 1.5TB worth of data on the N3700 box and it takes 70 hours.
I'm not familiar with the N-series storage, I didn't setup the backup job for them.
One thing I noticed is that Backup Exec sees the storage box listed under "active directory domains". Does that mean it sees the box as a server?
Since its agentless, what mechanism does it use for communication? the transfer rate is about 220MB/min, which is kinda low. Anybody has any clue to point me to the right direction?

thx.

Comments 7 CommentsJump to latest comment

teiva-boy's picture

 The IBM N series is a re-badged NetApp Filer.

You can back it up via UNC path, or NDMP.  The later being the fastest method.


There is an online portal, save yourself the long hold times. Create ticket online, then call in with ticket # in hand :-) http://mysupport.symantec.com "We backup data to restore, we don't backup data just to back it up."

BTSKE's picture

Thanks teiva-boy,
Could u give me some more insight?
what makes the NDMP protocol faster than UNC?
thx.

DominikG's picture

At the moment, the backup should be performed through another machine, on which the storage is mounted.
So the actual way is from the filer through a conencted server to the storage location (e.g. tape drive attached to the server where the filer is attached)
And if this tape device is only attached to the be server, the backup has to go through the BE server onto the tape.
With NDMP you are able to directly run the backup from the filer.
This for example lets you do a backup right from the filer onto a direct attached tape drive or through SAN.
Needless to say, that this should increase the backup time a lot...

best regards

Dominik

Consultant
PMCS.helpLine Software Gruppe
www.pmcs.de

follow us on Twitter: PMCS_NBU ; PMCS_BE ; PMCS_EV

BTSKE's picture

hhm.. i think I understand that much.
And your assumption is correct, i think the data goes to the BE server and then to tape.
But even this doesn't justify the rate at which it is backing up now. 5 hours for 4 gig, and then 70 hours for 1.2 TB, with an average of about 200MB/min. that just doesn't make sense. Something's off.
Its not the network bottleneck becasue when i do a copy and paste from the filer to the BE server, it comes through real quick, about 50 sec for 1 gig.
Its not the tape device because it does a decent job for other servers.
its something between BE and the N3700 box or some configuration. Any idea??

thx.

teiva-boy's picture

in 11d, the only NDMP that was supported was I believe a SCSI attached tape on the back of the filer.  It couldnt be used for any other backup task, only backing up that filer.  BE will send NDMP commands to the filer to move volume X to SCSI LUN Y, then talk back to BE when it's done.

In 12.5 you have more options for deployment.  remote, three-way, and shared tape.

NDMP is faster as much of the file system is ignored compared to NTFS based backups, as a steady stream of data is sent to tape ignoring the filesystem overhead.


There is an online portal, save yourself the long hold times. Create ticket online, then call in with ticket # in hand :-) http://mysupport.symantec.com "We backup data to restore, we don't backup data just to back it up."

DominikG's picture

Please also take care of the file count and file structure.
Backup Exec first builds the whole structure, before actually writing the data.
This consumes a big amount of time, and with a slow connection gets worse.
So not only the MB/GB are relevant.
I would always recommend doing NDMP backups from the filer to a directly / fast connected target device.

Consultant
PMCS.helpLine Software Gruppe
www.pmcs.de

follow us on Twitter: PMCS_NBU ; PMCS_BE ; PMCS_EV

BTSKE's picture

</backup_type><start_time>Backup started on 11/21/2009 at 8:32:20 AM.
</start_time><end_time>Backup completed on 11/21/2009 at 1:44:36 PM.
</end_time><summary><misc>Backed up 291722 files in 44555 directories.
</misc><new_processed_bytes>Processed 2,149,187,460 bytes in  5 hours,  12 minutes, and  16 seconds.

Does this make sense? 2GB in 5+ hours? in my environment this would take 20 mins or 30 mins max. And about 300,000 files is not that much at all.
And the other job is about 2 million files, 1.3 TB and takes 70 hours.

I know the NDMP method is all nice, but its outside the scope of this problem. so lets leave that out of this conversation.
I guess my objective (and the customer's objective) is to find the the bottleneck, if there's any, or any misconfiguration that could cause this. Because onviously something is not rite.