cancel
Showing results for 
Search instead for 
Did you mean: 

BE2010 - CASO; target duplicate-to-tape job to remote MMS

davei
Level 4
Hi all

Datacentre 1:
 
1x BE2010 (x64) - CASO installed
1x Quantum DXi 6500 OST appliance
 
Datacentre 2:
 
1x BE2010 (x64) - MMS
1x Quantum DXi 6500 OST appliance
1x HP 8-slot LTO4 autoloader (SAS)

I'm trying to implement what i thought was a disk-to-disk-to-tape backup setup, but might actually be disk-to-disk-to-disk-to-tape :)
 
I am using BE2010 to back up a number of things from vmware vsphere to simple windows servers and exchange\sql.  Backups working fine.
 
I'm using an OST disk-based backup device (Quantum DXi 6500) and, after no inconsiderable amount of messing, is sort of working as expected.
 
I have two datacentres that I am duplicating the backup sets between, each DC has a DXi 6500 device and a BE 2010 media server.  Duplication between the OST devices works fine.
I have the CASO option installed on BE server in DC1, and the DC2 media server reports in fine as a managed media server.
 
So my policies back up, for example, our vmware estate off the SANs, and plonk them on the DXi device.  Then a duplicate template kicks in and performs OST optimised duplication to the DXi device in datacentre 2.  Works fine and is actually impressively quick over a dedicated 30Mbps line.
 
What I would like to do now that those backup sets are in DC2 is to duplicate them again to tape locally in DC2.  DC2 has a BE2010 media server, connected to the replica DXi device, and also has an HP autoloader directly attached via SAS.  I would like to make the DC2 BE media server read the replicated media sets from Quantum DXi, and duplicate to tape in DC2.
 
I have found that CASO would 'load balance' my DC1 jobs without asking, so the other day i found BE MMS in DC2 trying to back up vmware in DC1 even though no connectivity to DC1 SAN!  Yes was running incredibly slow over the WAN in nbd mode.  So i 'fixed' that by restricting the selection lists to only run on the media server in DC1 - but i can already see that I think i need that selection list to be available to MMS in DC2 so it can duplciate to tape?
 
Enough waffling, the direct question i guess is:
 
Can anybody advise on configuration to achieve a duplicate to tape job on another site?  The obvious answer is to add another duplicate backup sets template to the relevant policy (using the previous duplicate template as the source for the second duplicate template?) but when creating this template i can't see how to force the job to run on a particular media server?
 
I have of course tried the above, but the second duplicate job which should go to tape just sits in a 'Ready' state (i told the template to use the tape device in DC2 but job doesn't start).
All the Quantum DXi LSUs are 'shared' in backup exec so both media severs can access them.
 
Any thoughts\comments would be great.
 
Thanks.
 
Dave
6 REPLIES 6

davei
Level 4
Further to my post above here is a breakdown of the test policy i am using:

1. Backup Template

Template Name:  Backup Template1
Source: local file on CASO server
Destination:  TestLSU1 on DC1 Quantum

2.  Duplicate Job 1

Template Name:  Duplicate Template1
Source Template:  Backup Template1
Destination:  TestLSU1 on DC2 Quantum
Preferred Source:  TestLSU1 on DC1 Quantum

(the above two steps work great)

3. Duplicate Job 2 (to tape)

Template Name:  Duplicate-to-tape1
Source Template:  Duplicate Template1
Destination:  HP1 (tape) on DC2 MMS
Preferred Source:  TestLSU1 on DC2 Quantum

If there is any more information i can provide that might help let me know!

Thanks in advance for any ideas.

Cheers

dave

Ray_Ebanks
Level 3
Davei, did you find a solution to your implementation?  I'm implementing a similar solution (disk-to-disk-tape) to (disk-to-disk-to-tape) at remote site but I’m unable to get the duplicate jobs working on the managed media server between main site and remote. Can you share what you have gathered so far?

davei
Level 4
Hi Ray

I may be way off the mark with this but in the absence of any other more specific documentation or outside help, I think that my problem is catalog replication between a CAS and an MMS.  It seems to me that an MMS replicates it's catalogs to its CAS, but a CAS does not replicate it's catalogs 'downstream' to an MMS.  This meant that the MMS with the tape drive attached did not seem to be aware of operations that had gone on on the CAS. 

I have (and had) my MMS catalog setting configured to 'replicated' under the MMS properties in the 'Media Servers' tab on the CAS.

What i am in the middle of doing is swapping the roles of the CAS\MMS servers - i now have my CAS in our DR site (with the tape-drive locally attached to it), and the MMS up in our production datacenter. 

As it happens I am waiting on the first disk-disk-(disk?)-tape jobs to complete at the moment.  Run into a few difficulties with catalogs etc. since swapping the roles but am working through that (eg. my first duplicate job is sat in a 'loading media' state right now as it seemingly can't find the media it wrote to 30 mins ago.....).

My theory is that when the backup template completes on the MMS, it will replicate the relevant catalog changes to the CAS.  Then when the disk-disk duplicate template completes (which due to Openstorage optimised duplication can run on either the CAS or the MMS), it will also update the catalogs on the CAS to show the duplication occurred successfully.  Hopefully then, the CAS will be aware of all the operations that have gone on, and will be able to duplicate to tape locally in DC2.

Regarding targetting the job to the MMS, i found that in the properties of the relevant selection list, on the 'device' page the option 'Restrict backup of the selection list to devices on the followiing medai server or media servers in a pool' enabled me to ensure that the backup template\job associated with that selection list always runs on the MMS rather than the CAS.

With any luck i will know more later on today so will update when i can.  Have two weeks leave planned so bad timing really!

Good luck.

Dave

Colin_Weaver
Moderator
Moderator
Employee Accredited Certified
I would be interested in your results

Ray_Ebanks
Level 3
Davei , i think you are on the right track. 

Davi >>>>I have (and had) my MMS catalog setting configured to 'replicated' under the MMS properties in the 'Media Servers' tab on the CAS.

There are 3 available options for catalog settings..(Distributed, centralized and replicated). My MMS was set to distributed...is now set to replicated.

Davi >>>my first duplicate job is sat in a 'loading media' state right now as it seemingly can't find the media it wrote to 30 mins ago.....).

You may want to check the over write protection period on the Tape since you had written to it 30 minutes ago. A quick erase will fix that issue.

I was able to get my jobs running on the MMS as duplicate policy enabled jobs. However, these jobs did not work when I use the duplicate existing backup set option since I was unable to delegate MMS media location while on the CAS.  

davei
Level 4
Hi

Back from leave. 

Ray - i now have the same problem.  Backup jobs delegate fine to the MMS and run as expected - but i can't force the duplicate job to run on the MMS, it keeps trying to run on the CAS.

From BE 2010 Admin guide, page 1501 "Requirements for duplicate backup data and synthetic backup jobs in CASO":

"A recurring job from a policy that contains a Duplicate Backup Sets template or a Synthetic Backup template must be run on the same managed media server where the job was initially run."

Ok that bit makes sense right (other than providing no way to force the Duplicate template to run on a specified MMS)?  But then the next sentence:

"The jobs produced from these templates require access to the media that contains the backup sets that were produced from the preceding jobs".

Well, in my particular case I am using shared storage media (the OST disk based dedupe devices), and both CAS and MMS have access to and successfully mount the LSUs on both OST devices.  So both servers DO have access to ALL media created by eachother.

So in my scenario, in the absence of a way of forcing the Duplicate template to run on the MMS rather than the CAS, I would have thought that my CAS would just connect to the LSU in the production datacenter and initiate the OST optimised replication.  But it doesn't.

Ray - you say you were able to get jobs running on the MMS as duplicate policy enabled jobs - sounds like what I am after?  How did you achieve this?

Thanks.

Dave