Video Screencast Help

File System Archiving in Enterprise Vault v9: Placeholder Migration

Created: 08 Sep 2010 • Updated: 16 Sep 2010 • 15 comments
Darren Locke's picture
+4 4 Votes
Login to vote

Over the coming weeks I'm going to be writing a series of blogs detailing the enhancements to Enterprise Vault File System Archiving in v9. While these may not be big headline grabbing features, I think our existing customers will find them extremely valuable and beneficial. After all, the enhancements that we have delivered for FSA in v9 have all been directly driven by customer requests and feedback.
 

I'm going to start off by talking about placeholder migration. Once you have deployed FSA, the ability to easily manage the archived content and to be able to move placeholders around (without recall) for things like hardware refreshes or share re-organisations is key. Now some of you may be thinking that doesn't FSAUtility already have the ability to move placeholders? The answer is yes it does, but it has specific use cases associated with it, which are not appropriate for some of the more common migration tasks.
 

Let me explain more. The exiting FSAUtility -m option will move placeholders from one volume to another, from one server to another server or even across disparate platforms (e.g. Windows to NetApp) but it assumes that the source and target destinations are both being managed by different Enterprise Vault servers. What this means is that along with moving the placeholders, we also move the underlying archived files in the backend. So all archived files are exported from the source archive and imported in to a new archive on the destination.
 

What this means is that the speed of the migration is determined by the size of the archived files within Enterprise Vault, not by the number of placeholders on the file system. You need to use this when your destination file server is being managed by a different Enterprise Vault server to that of the source file server. But what if the source and destination file servers are both being managed by the same Enterprise Vault server?
 

FSAUtility -m can still be used in this scenario, but your migration times are going to be long. You asked the question to us about why do you have to move the backend data if it does not have to move across Enterprise Vault servers? It's a very good question and now with the FSAUtility -pm option you no longer have to move the backend data.
 

So, providing source and destination file servers (and they can be disparate devices) are being managed by the same Enterprise Vault server you can use -pm for the migration. Here the time to complete will be determined by the number of placeholders on the volume not by the size of the archived files they point to. This will mean you have a highly performant placeholder migration tool. Often you have to do server migrations, volume re-organisations etc. in small maintenance time-windows. You now have the tools from us to help you succeed in those migrations.
 

FSAUtility will still only move the placeholders whether you use the old -m or the new -pm option. We recommend using something like robocopy for the non-archived files. Robocopy is a great tool for this as it has the /XA:O option which allows it to skip files with the offline attribute set i.e. our placeholders.
 

FSAUtility -pm will move whole archive points (archives) from one volume to another, including all placeholders below the archive point. If you want to move part of an archive point you will need to use -m as you are effectively splitting an archive.
 

In our labs we have seen a very encouraging number of placeholders be migrated per hour. I would be interested in hearing your experiences with the new -pm option and the type of migration throughputs you are getting in your environments.
 

For more information on how to use FSAUtility, check out the 'Utilities' guide in the documentation folder of the v9 kit.
 

Over the coming weeks I will be giving more details on the following enhancements for FSA in v9:
 

  • FSAUtility: bulk recall
  • FSAUndelete tool
  • FSA Reporting performance and scaleability improvements
  • FSA Checkpointing improvements
     

So keep coming back for those or subscribe to the RSS feed. I would also be interested in hearing what other topics around FSA you would like us to blog about.
 

Darren

Comments 15 CommentsJump to latest comment

Maverik's picture

Excellent blog and useful feature.  Thanks Darren.

+1
Login to vote
spin99's picture

Thank you for the info.  Any chance that we can use FSAUtility version 9 with version 8 of EV?  Our performance is really slow and this would help out quite a bit.

 

Thanks,

 

Chris

-2
Login to vote
Darren Locke's picture

Hi Chris, Unfortunately you will need to upgrade to EV9 to take advantage of the new -pm options.

Darren

-6
Login to vote
Dataway's picture

Can we personalise a "Timeout message for File System Archiving"  for files which are archived and migrated to secondary storage using NBU i.e. Tape once recalled by a user?

-4
Login to vote
Darren Locke's picture

Hi Dataway, 

The timeout message that you get is an operating system generated message. Unfortunately we are unable to change that without having some form of client software on the users' workstation. So it's down to user education at the moment. They need to understand that while they may get the timeout message, the file will still be recalling in the background and will come back to disk. 

Darren

+2
Login to vote
RichardF14's picture

Hi Darren,

 

Will this also work if there are two Enterprise Vault server each inside a different domain.

There is a two-way trust between both domains.

 

Richard.

 

0
Login to vote
Darren Locke's picture

If the EV servers are part of the same EV installation, then with the two-way trust in place, this should work. If the EV servers are in disparate EV installations, then this will not work.

Darren

0
Login to vote
Sergy Stouk's picture

Thanks for the blog.

I have been managing financial sector business data for multiple departments for the last several years and we went through many data migrations between the servers which vaulted files in them.

The option and the ability to move shortcuts is pretty much useless to us as we never "move" neither data nor shortcuts.

We have to make sure the old and new environments are identical and have completly synchronized  data with permissions in both locations, accessible and tested for several days at least by business representatives each time such migrations are done - therefore we have to "copy" the data first, test it in new location, gather sign-offs and only then redirect end-users to use it.

"moving" data  or shortcuts using any utility will disrupt end-users and will take days as we are talking about  TBs of data with millions of files per business department.

In most cases we simply restore the data (including the shortcuts) from latest tape backup, make sure shortcuts are accessible from both servers and then go on with the rest of the plan.

If FSAUtility would had an option for "duplicating" the shortcuts in a new location - that would had been a much more anticipated feature in our example.

0
Login to vote
Darren Locke's picture

Sergy - thanks for your comments. The shortcut duplication feature is something we have heard before, but the biggest demand we had received was for the placeholder movement options we have delivered. The speed of movement will be dependant on the number of placeholders and the size of the folder structure. In our labs we have moved 100'000s+ of placeholders per hour. So while there is disruption we would hope that the move operation (for placeholders at least) could take place during a weekend maintenance window. Give it a try on a test folder structure and see how you get on.

Darren

-2
Login to vote
teshp's picture

Hi Darren,

I am working on a migration and trying to work out how you guys got 100K per hour migration of place holders.

Environment:

W2K8 R2 Server with 2TB Volume part of FSA V9 Environment. With own dedicated SQL server.

The server will be migrated to EMC Celerra part of the same FSA V9 Environment.

All on own dedicated 1GB network, all user traffic on segregated network.

 

However when I tried to run the –pm switch on the test environment with only getting around 20K per hour migration of place holders.

 

Could you give more details of how you managed 100K per hour.

 

Thanks

0
Login to vote
dksoft's picture

Good day.

What's heppen with Archive Explorer Tree? Do FSAUtils update this structure after placeholder move?

Thank you

0
Login to vote
LOXII's picture

Thanks for the blog - very useful

 

I am also working on share re-organization, Joiner/Mover/Leaver process and other similar process within an EV FSA infra.

 

The FSAUtility –pm option is working fine for the move. I arrive to a rate of 35 K files/hour in our test environment.

And a robocopy /XA:O achieves the job.

 

But it will imply a disruption of the service (you mention during the weekend).

While a copy can be done during the working hours and at launch moment, apply just an incremental before doing the change will imply a disruption of a few minutes.

 

I am wondering if there another way to perform a copy of files shares without recalling all the archived files.

If anybody have information’s, reference or whatever concerning the Joiner/Mover/Leaver process within a SEV FSA infrastructure, thanks for letting us know.

 

Thank you

-2
Login to vote
AWMorris's picture

Hi Darren.  I was wondering if you had any comments for the following link?

 

https://www-secure.symantec.com/connect/forums/fsautility-and-checkpoint

0
Login to vote
ngenmercedes's picture

Excellent post! We ran the command, but get an error message:

Error: There are no archives under the source path \\server\path that can be migrated

I can't find any reason why. Any thoughts?

0
Login to vote
ia01's picture

Excellent Post! Thanks Darren!

Is there any improvements or new featues on FSAUtility for EV 10.0.1 ?

0
Login to vote