cancel
Showing results for 
Search instead for 
Did you mean: 

Backup Exec Deduplication - Seeding and 1st Backups

BackupExecNewb
Level 4

Hi,

I have a number of sites that I want to use the BE2014 dedup option for my backups.  The remote sites have 2TB plus of data.  It will take forever to back this up over the wire.  Symantec had recommended running up a media server at the remote site, backing up the data to a dedup volume, and then move the BE server and dedup volume to the local site and run a duplicaiton job.

Has anyone had any expeience with this?  Ideally I'd like to use a VM to act as the temp remote site media server.  However, storage will still be an issue, as I don't have enough local storage to present to the ESXI host.  USB drives are not supported for dedup volumes, so all I can think of is a cheap NAS box running iSCSI.  Ideally, I'd want something that would be easy to transport around. 

Also I imagine I'll need CASO in order to duplicate the dedup volumes between media servers.  This isn't available with the trial version of BE is it?

Most importanrly, I'd also like to verify if this method is carried out, that the subsequent 1st full backup from the remote site to the duplicated dedup volume will in fact run quickly, as the data should now be seeded.

I'd love to hear from anyone that has had some real world experience with this.  I would hate to ship it out, set it up, only to find out that it doesn't work.

 

Thanks

2 ACCEPTED SOLUTIONS

Accepted Solutions

CraigV
Moderator
Moderator
Partner    VIP    Accredited

Hi,

 

CASO is bundled with the Enterprise Server Option and I don't think it is available for trial, but run through the licensing wizard and check to confirm.
Read the TN below on how to seed your first backups:

http://www.symantec.com/business/support/index?page=content&id=TECH163110

Thanks!

View solution in original post

pkh
Moderator
Moderator
   VIP    Certified

No.  Not sending data across the wire may not reduce the duration of the backup significantly and while the job is running, you risk job failure.  Doing a local backup reduces the chance of failure due to poor link quality.

View solution in original post

18 REPLIES 18

CraigV
Moderator
Moderator
Partner    VIP    Accredited

Hi,

 

CASO is bundled with the Enterprise Server Option and I don't think it is available for trial, but run through the licensing wizard and check to confirm.
Read the TN below on how to seed your first backups:

http://www.symantec.com/business/support/index?page=content&id=TECH163110

Thanks!

BackupExecNewb
Level 4

Thanks - useful information.

I have the remote site data close to hand though via dfsr replication.  I can seed with that, but it is still taking a considerable amount of time to perform the 1st full remote site backup.  I get maybe 600GB in and the job fell over with an error code E008488.  Was told this could be several things but likely a network glitch. I can't afford to wait 2 days into my first backup for it to fail and then I have to start again.

I guess what I'm really trying to find out if the method outlined above in the TN would speed up that intial backup from the remote site, or if there's anything can do to adress that 1st remote site backup after the data is seeded.

CraigV
Moderator
Moderator
Partner    VIP    Accredited

...by seeding your first backup to local storage and then transporting to your central site, you're going to save your bandwidth. It should go through, but don't forget that your first backup is always going to be your longest. Just because you're using dedupe doesn't mean it's fast from the get-go.

Subsequent backups will be faster as more redundant data is found and less is backed up.

Thanks!

pkh
Moderator
Moderator
   VIP    Certified
Are you going to setup a media server with dedup at each of the remote sites? If you are thinking of using your central site to backup your remote sites, think again. I doubt you have the bandwidth

BackupExecNewb
Level 4

Totally understand. I just thought it would go faster when the data isseeded.  I did a 1st backup of remote site data with just 1GB of seeded data and it took 6 minutes on a 20 Mbps link. Given those calculations, I estimate 7.3 days to do 2TB.  I can live with that, but not if it fails 2 days in and I have to start again.  I don't even want to think about how long it will take to backup the 2TB site with the 6Mbps link.

Perhaps this is where optimised duplication beats just seeding and backing up??

BackupExecNewb
Level 4

No I'm wanting to use BE dedup exclusively to back up the remote sites from one server at the data center. (when the data is seeded).   That's what the technical white paper says the dedup option can do; to use client side deduplicaiton to reduce the amount of data sent over the wire.

CraigV
Moderator
Moderator
Partner    VIP    Accredited

...mmm, thought that you had storage on the local site and wanted to do optimised dedupe...which is what would be better suited for this setup.

Thanks!

BackupExecNewb
Level 4

We're trying to reduce costs by not having media servers at each site.

CraigV
Moderator
Moderator
Partner    VIP    Accredited

...then backing up across the WAN is the only other option. Also, seeding in this case should do the trick.

If you can test this, then do so before committing to it.

Thanks!

BackupExecNewb
Level 4

OK - just to confirm:

There's no difference between seeding and running the first backup and optimized dedup for seeding and then the first backup?  Either way, the remote server will still take the same amount of time to backup, as it's got to go through to know that the data s already there?

 

pkh
Moderator
Moderator
   VIP    Certified

How are you going to do an optimised duplication without a dedup folder at the remote location?

Backing up your remote servers from the central site means that if there are any dropped packets between the two sites, your job will fail.  This is regardless of whether you do any seeding or using client-side dedup.

BackupExecNewb
Level 4

I'm not.  I'm trying to determine whether it's worth temporarily setting up an optimized dedup environment to expedite the 1st backup process.  Untilmtately, I will be only be using remote site backups to a central site dedup volume.

I can already seed the remote site data pretty easily,as I have backed up the remote site DFSR replica data to the dedup volume.  Small remote site 1st backup tests (i.e. 1GB) indicate that not much data's going over the link, which means client side dedup is working. However, when I do this with 500GB - 2TB of data, it going to take a very long time.  And the longer it takes the greater the chance of the job failing due to dropped packets.

If temporarily setting up another media server and running optimized duplication betweeen the sites would make the initial backup go faster and/or be more reliable then it would be worth the effort.  If, however, it still will take the same amount of time to do the first initial backup, regardless of how the data is seeded, then there's no point, and I'll just have to let the 1st full backup run and hope for no failures due to dropped packets.

pkh
Moderator
Moderator
   VIP    Certified

It is not just the first full backup that you have to worry about.  You would have to do full backups often so that your backup chain does not become unwieldy.  You cannot do incrementals forever.

BackupExecNewb
Level 4

Absolutely.  But isn't the whole thing that this shoud not take long if the data is unchanged?  The majority of the data is stagnent.  Even a full shoud not result in the data being copied over the wire.

pkh
Moderator
Moderator
   VIP    Certified

No.  Not sending data across the wire may not reduce the duration of the backup significantly and while the job is running, you risk job failure.  Doing a local backup reduces the chance of failure due to poor link quality.

teiva-boy
Level 6

BackupExec's Client-side dedupe, does virtually nothing to speed up backups.  All it does is reduce bandwidth.  I've found in almost every case the backup durations are roughly the same.

That said, perhaps the Checkpoint restart option will be your best bet?  http://www.symantec.com/business/support/index?page=content&id=HOWTO99495

You'll still get failed jobs, but it'll hopefully try to restart where it left off.

 

If you get tired of this, There is always EMC's Avamar, that pretty much blows the doors off anything for WAN based backups, but it'll cost ya...  The old Symantec PureDisk actually did okay with remote WAN backups, but they killed that product after migrating the core dedupe engine into BE and NBU.  Then didn't bother to port some of PureDisks key features that made it good for remote WAN backups...

BackupExecNewb
Level 4

thanks for the feedback

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

More than a month old but this posts is popping up under 'Can you solve these?'

Herewith replies to my question about the same topic:

Seeding before Dedupe Backup across WAN