Video Screencast Help

Seeding before Dedupe Backup across WAN

Created: 26 Nov 2012 • Updated: 27 Nov 2012 | 11 comments
Marianne's picture
This issue has been solved. See solution.

We are proposing 3 x 3600’s for a customer to be placed at different sites with CAS at H/O.

There was a question about a single client that needs to be backed up in a remote site (let's call it Site4). 

The idea is to back it up across the WAN to Site1, but we need to understand how to seed the first backup.

I have found a couple of forum posts as well as TN’s, but still unable to figure out a short ‘howto’ feedback to the customer:

https://www-secure.symantec.com/connect/forums/backup-exec-2012-optimized-deduplication

https://www-secure.symantec.com/connect/forums/pre-stage-data-slow-wan-deduplication

I like Colin Weaver’s reply here: https://www-secure.symantec.com/connect/forums/pre-stage-data-slow-wan-deduplication#comment-3953971 but fail to understand the HOWTO part.

https://www-secure.symantec.com/connect/forums/how-seed-baseline-full-backup-synthetic-backups

http://www.symantec.com/business/support/index?page=content&id=TECH162822

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

Advice gathered so far can be summarized as follows:

1. Back up your data to disk or to tape.
2. Transport the tape or disk to the destination system.
3. Inventory and catalog the data.
4. Create a duplicate backup job from the disk or the tape to the deduplication storage folder.

 

Seems this will only work if there is a media server at the remote site, right?
The remote site is way too small to warrant a seperate media server license.

How do we perform this first backup to tape or disk when there is no media server at the remote site?

Comments 11 CommentsJump to latest comment

jandersen's picture

Hello Marianne,

Here is a simple approach:

Given it is plain filesystem data you just have to copy the data for the client at the remote site to an external USB storage.
Then you transfer the USB drive to you local backup environment and back it up through a local client to the dedupe pool.

When the first backup of the client takes place fingerprints are calculated at the client side, but there is now need to transfer the blocks over the WAN as they are already known to the dedupe store.

--jakob;

Marianne's picture

Thanks for the suggestion, Jakob, but my concern is about Rafael's experience in the first post that I've mentioned above.

See https://www-secure.symantec.com/connect/forums/pre... and https://www-secure.symantec.com/connect/forums/pre...

Supporting Storage Foundation and VCS on Unix and Windows as well as NetBackup on Unix and Windows
Handy NBU Links

jandersen's picture

Sorry,

Missed that point.

I have done it with MSDP on NetBackup between MSDP's in Hina and Europe. I can't see (at a high level) what the difference should be with Backup Exec.

Anyway. On NetBackup you solve the problem with

http://www.symantec.com/business/support/index?pag...

But I haven't found the equivalent feature for BE (I'm not a frequent BE user)..

 

jandersen's picture

Hmm,
Not sure if I'm double posting now - my recent post got lost (at least I can't see it myself).
Anyway I'll try again.
I have used http://www.symantec.com/business/support/index?page=content&id=TECH144437 for NetBackup.
I hope for that there is an equivalent in BE (I'm not a frequent BE user).

--jakob;

 

 

pkh's picture

How do we perform this first backup to tape or disk when there is no media server at the remote site?

You can't.  The advice is based on being able to do the first backup at the remote site.

What you can do is to bring one of the appliance to this remote site and do a backup.  Bring it back to the main site and then do an optimised duplication to the main site's dedup folder.  After this, you can deploy the appliance to the other sites.

SOLUTION
Mark_Solutions's picture

Marianne

Not sure if you can do a similar thing to in NetBackup as I am not totally familiar with the Backup Exec de-dupe engine but it does appear that it does use the pd.conf on the client.

The principle is to back up a similar client locally first and then edit the pd.conf of the remote client to include the client and job name and date of the local client you have backed up

This seeds the fingerprint cache and tells the remote client to use that cache to minimise the amount of data it has to pass over the network.

I would imagine this is possible as the engines behind it are the same so it would be down to exactly what you put in the pd.conf file to make this work with Backup Exec. In NetBackup it is: clienthostmachine,backuppolicy,date

The other option, if possible, is to locate the appliance at the remote site, take a backup of the remote client and then bring it back to the data centre - again down to logistics that one!

Hope this helps

 

Authorised Symantec Consultant

Don't forget to "Mark as Solution" if someones advice has solved your issue - and please bring back the Thumbs Up!!.

Marianne's picture

Thanks guys!

Jakob - for some or reason the spam filtering software has quarantined your last two posts. I have 'released' them.

Thanks for the NBU advice, gents - yes, I am also aware of the NBU way of doing this. Because of our long-time experience with this product, we sort of expect to find the same features in BE.

I like pkh's idea - as long as I will be the one to fly down with the appliance!

This advice is sort-of in line with advice from our local TSE (who finally responded to my request):

The 4 steps in my opening post can be accomplished with backing up to a laptop with BE Server software installed (using Demo license) and external USB disk.

Would you agree with this advice?

The appliance taken to remote site seems like the least amount of hassle...

Supporting Storage Foundation and VCS on Unix and Windows as well as NetBackup on Unix and Windows
Handy NBU Links

Mark_Solutions's picture

Marianne

The four steps sound good but I would keep the backup job as identical as possible to the one you setup on the live site afterwards

If it does work like NetBackup then it seems to take the job name into account when fingerprinting as well as the client name so that may get the best out of it.

Perhaps as you are using CASO you could add the laptop as a temporary MMS with a job that backs up to its external USB and then duplicates to the Appliance (with the duplication submitted on hold)

Ship the laptop and USB drive to the remote site to run the backup, ship it back and make sure it is up and running and then take the duplication off hold so that it duplicates the same job / client etc. to the Appliance.

That gives you a "real" job for that client which can then be edited to set it to be backed up by the Appliance from then onwards - as identical a setup as possible?

Hope this helps

Authorised Symantec Consultant

Don't forget to "Mark as Solution" if someones advice has solved your issue - and please bring back the Thumbs Up!!.

Marianne's picture

Thanks Mark.

Yes, CASO will be deployed in this environment with laptop as temp MMS. 
steps outlined in your post makes perfect sense.

Taking the (new) 3600 to remote site also seems to be a good idea.

I will discuss with colleagues and let you know what our final recommendation to the customer will be...

Supporting Storage Foundation and VCS on Unix and Windows as well as NetBackup on Unix and Windows
Handy NBU Links

pkh's picture

The BE dedup option also uses the pd.conf file since it is derived from Puredisk.  However, the implementation in BE is different from that of NBU and I am not aware of the parameters that Mark suggested.

However, you don't have to do what Mark has suggested to seed your dedup folder.  To the dedup engine, a chunk of data is a chunk of data no matter which job or server it comes from.  For example, if data Chunk A is written by Job1 from ServerA and Job2 from ServerB happens to have the same chunk, then this chunk from Job2 is not stored.  The idea of seeding the dedup folder with a backup from that particular remote server is so that you can get as much similiar chunks into your dedup folder as possible.  If you have another server with similiar data as the remote server (rare), you can actually use that to seed your dedup folder.

Between carrying the laptop or the appliance to the remote site, I would opt for the appliance although it is bulkier.  This is because you can be sure that it is working.  If you take a laptop and BE is not set up properly, you may end up fumbling with BE at the remote site.  Not a very comfortable feeling.

Marianne's picture

Thanks to everyone for your thoughts and suggestions!

Appliance will be option no. 1. 

Supporting Storage Foundation and VCS on Unix and Windows as well as NetBackup on Unix and Windows
Handy NBU Links