VBR Backup Copy for VEB backups

[Fledge]

Hi,

Think I'm confusing myself the more I look into this, so reaching out for some sanity...

If using VBR as the repository for VEB backups
And using VBR Backup Copy to copy the VEB backups to a VBR repository at a remote site
What does VBR Backup Copy actually do with the a local VEB backup set ?

Ok the first 'copy' will copy the 'full' across to the remote site data mover/repository
But subsequently does VBR Backup Copy actually read the local repository VEB backup chain and then transfer across to the remote site data mover effectively 'incremental' updates
And the remote site data mover/repository does the smarts to make its own full + incremental chain ?

If it does what happens when the VEB backup in the local repository hits its retention period?
i.e. the VEB backup (forward forever) does synthetic update to its full to drop incremental's from its chain
Would VBR Backup Copy be forced to transfer a full across to the remote site data mover/repository

Thanks in advance for any brain ache help !

[jmmarton]

It's important to understand how backup copy jobs work. The key is from this section of the User Guide.

https://helpcenter.veeam.com/docs/backu ... tml?ver=95

Veeam Backup & Replication copies VM data per VM at the block level. That is, it does not copy the whole VBK, VIB or VRB files from the source to target backup repository. Instead, it works with data of separate VMs stored in these files.

When using VEB, it's a forever forward incremental. All that lands in the VBR repo is changed blocks. When that's used as a source for a BCJ, only those changed blocks are transferred. That doesn't change when the retention period is hit and there's a merge/transform of the full backup, as all the BCJ is looking for are new changed blocks.

Joe

[Fledge]

Hi Joe,

Thanks very much for the quick reply - the last paragraph is exactly the clarity I was looking for. Had previously started at the url you mention but in a journey through other doco and forum postings I ended up in confusion around the case specific to VEB backup sets (as these not VBR VM backups, rather host agent based file level backups, wasn't sure of 'CBT' in this context).

For example it says at https://helpcenter.veeam.com/endpoint/1 ... _copy.html

"Backup copy jobs processing Veeam Endpoint backups have one limitation: you cannot use backup mapping for Veeam Endpoint backups. As a result, if you have a full Veeam Endpoint backup on the target repository, you will not be able to use this backup as a "seed" for the backup copy job. The backup copy job will always copy the whole Veeam Endpoint backup chain to the target repository "

Which I understood to mean you couldn't seed a target repository but that once BCJ had copied the VEB full across it would subsequently only effectively send incremental's (not always try and copy the whole chain across every time) ?

The remote repository will be located across the other side of a WAN, we could point VEB directly to the remote repository (or just a remote share) to get an off-site backup set but VEB (understandably) has limitations and not perform well across a WAN (VAW may address some of the VEB functional limitations but am assuming it still would not be very WAN friendly).

BCJ would appear to be a much better fit architecturally and by assumption performance wise, hence why seeking confirmation that using VEB backup as the source for a BCJ it will always only transfer changed blocks across the WAN to the remote repository (i.e. inherently minimise the WAN traffic involved) - it being the remote repository that effectively handles the backup set at that end for say its merge / transform functions ?

For completeness (and to give others all the info in one forum hit) - what would be the minimum Veeam components required for a remote repository and what would the license implications be ? (where the remote repository's only role is to act as the destination for BCJ from the existing licensed on prem VBR)

Thanks and have a good weekend

[jmmarton]

Which I understood to mean you couldn't seed a target repository but that once BCJ had copied the VEB full across it would subsequently only effectively send incremental's (not always try and copy the whole chain across every time) ?

This means when creating a new BCJ using an Endpoint backup as the source, the entire chain is copied so you can't use an existing copy of VEB backups in the target to seed the BCJ. This doesn't mean the entire chain is copied on every run of the BCJ.

BCJ would appear to be a much better fit architecturally and by assumption performance wise, hence why seeking confirmation that using VEB backup as the source for a BCJ it will always only transfer changed blocks across the WAN to the remote repository (i.e. inherently minimise the WAN traffic involved) - it being the remote repository that effectively handles the backup set at that end for say its merge / transform functions ?

The type of repo will determine who handles merge/transform operations. If it's a Windows repository then yes the repo handles this directly. But for other repository types, such as CIFS repos, the gateway server handles this.

For completeness (and to give others all the info in one forum hit) - what would be the minimum Veeam components required for a remote repository and what would the license implications be ? (where the remote repository's only role is to act as the destination for BCJ from the existing licensed on prem VBR)

I'm not sure I completely understand the question. BCJ copies directly from a source repo to a destination repo. No other components are involved except for possibly WAN accelerators if those are configured in the BCJ.

Joe

[Fledge]

Thanks again Joe - the above clear now
Some context for you on the last point

At the remote location, installing a 'backup repository server' (on a windows host) will install the backup repository and the data mover components and that is everything required for BCJ between a full VBR set up at the main site and the remote site ? (excluding optional WAN accel)

The VEB backup we want to BCJ to the remote site is effectively the backup of a very large data drive and the remote site performs a DR role (the host that mounts the data drive is already backed up to the remote site through a separate process). If the main site goes away, will the VEB instance on the remote site host be able to see the remote site backup repository to let it use the VEB backup to restore from (i.e. in such a scenario VEB will be able to see the backup repository with only the above components installed) ?

Wasn't sure if any other Veeam infrastructure components might be needed in this scenario and if it was just a case of installing a 'backup repository server' does that attract any additional lic cost (or as we are just using BCJ function, that is included in the main site VBR lic) ?

Thanks again

[veremin]

At the remote location, installing a 'backup repository server' (on a windows host) will install the backup repository and the data mover components and that is everything required for BCJ between a full VBR set up at the main site and the remote site ? (excluding optional WAN accel)

Correct.

If the main site goes away, will the VEB instance on the remote site host be able to see the remote site backup repository to let it use the VEB backup to restore from (i.e. in such a scenario VEB will be able to see the backup repository with only the above components installed) ?

So, in case of disaster you will only need to restore one particular volume, not a machine as a whole (bare-metal recovery)?

Wasn't sure if any other Veeam infrastructure components might be needed in this scenario and if it was just a case of installing a 'backup repository server' does that attract any additional lic cost (or as we are just using BCJ function, that is included in the main site VBR lic) ?

Installing repository in remote site does not consume license.

[Fledge]

Hi,

So, in case of disaster you will only need to restore one particular volume, not a machine as a whole (bare-metal recovery)?

Yes that correct - there's some complexity to the scenario which ends up driving the setup described; the VM will already exist at the remote site (so no bare-metal recovery), instead VEB on that VM will be used to restore one particular volume (a large data drive)

[veremin]

You will need to install VEB, locate backup file and proceed with volume-level restore.

[Fledge]

We setup what was outlined above - so have the following food-chain in place

Main Site:
Data Host using VEB agent to backup a large data volume to a local VBR server/repository
BCJ on the local VBR server copying the VEB backup across WAN to a repository installed on a Windows server at the DR site

DR Site:
Windows repository server with just the Veeam repository service running
Repository has result of the Main site BCJ - i.e. a backup chain from the VEB agent backup at the Main site

All good and as expected so far - and BCJ plays nicely over the WAN

However
If trying to do a VEB restore from a host at the DR site, the VEB (either file level or volume level) restore is looking for a VBR server it can connect to to tell it the location of repositories
It does not appear able to talk directly to a repository ?
But as this is the DR site, there is no connection to the Main site VBR (as would be the case in a disaster)
There is only the Veeam repository service running on the Windows repository server at the DR site
So does this mean that we would also have to have VBR running at the DR site
Just so that VEB can talk to VBR for it to tell it that about a repository that it can then use ?

On the VEB agent have tried the following
If under OS I network mount the dir used by the repository and tell Veeam Endpoint Recovery that the backup is in 'local storage' it errors saying VEB cant open backups created by VBR
If in Veeam Endpoint Recovery I select instead 'network storage'
Then select 'shared folder' it ends up with the same error (not surprising)
If instead select 'Veeam backup repository'
It then asks for the IP for the Veeam backup server (as in 'specify VBR server to retrieve the list of backup repositories from')
The Windows repository server is only running the Veaeam repository service, so pointing to this end with error cannot connect with VBR server

Have we missed something?

[Fledge]

Posting this to a new thread in case it gets buried in an old one

We did some initial setup checking that was covered in the following forum thread

veeam-backup-replication-f2/vbr-backup- ... 41622.html

The food chain we have in place is as follows

Main Site:
Data Host using VEB agent to backup a large data volume to a local VBR server/repository
BCJ on the local VBR server copying the VEB backup across WAN to a Veeam repository installed on a Windows server at the DR site

DR Site:
Windows repository server with just the Veeam repository service running
Repository has result of the Main site BCJ - i.e. a backup chain from the VEB agent backup at the Main site

All good and as expected so far - and BCJ plays nicely over the WAN

However
If trying to do a VEB restore from a host at the DR site, the VEB (either file level or volume level) restore is looking for a VBR server it can connect to to tell it the location of repositories
It does not appear able to talk directly to a repository ?
But as this is the DR site, there is no connection to the Main site VBR (as would be the case in a disaster)
There is only the Veeam repository service running on the Windows repository server at the DR site
So does this mean that we would also have to have VBR running at the DR site
Just so that VEB can talk to VBR for it to tell it about a repository that it can then use ?

On the VEB agent have tried the following
If under OS I network mount the dir used by the repository and tell Veeam Endpoint Recovery that the backup is in 'local storage' it errors saying VEB cant open backups created by VBR
If in Veeam Endpoint Recovery I select instead 'network storage'
Then select 'shared folder' it ends up with the same error (not surprising)
If instead select 'Veeam backup repository'
It then asks for the IP for the Veeam backup server (as in 'specify VBR server to retrieve the list of backup repositories from')
The Windows repository server is only running the Veaeam repository service, so pointing to this end with error cannot connect with VBR server

Have we missed something?
Or does this mean that we also have to have VBR running at the DR site, so that VEB can talk to VBR for it to tell VEB about a repository ?
(and that would be VBR's only role at the DR site - i.e. to provision for a VEB recovery from a repository, VBR would be performing no backup functions)

[Fledge]

For anyone else following this path...

VEB cannot access directly a backup chain in a repository on a server that runs only the repository/data mover components

On the DR site you have to have a VBR install to point a VEB agent at for any VLR or FLR operations

The DR site VBR does nothing else instead of act as broker between the VEB agent on the DR host and the DR repository containing the BCJ backup chain from the main site

The repository at the DR site is written to and updated by the BCJ at the main site (so that the DR site repository contains the latest backup from the main site VEB agent)

You add the DR site repository to the DR site VBR (making sure you select import existing and import index at the review page last step)

The DR site VBR does not appear to do anything with the DR site repository you add (the main site VBR BCJ keeps writing the backup chain in it)

In a DR scenario (main site gone)
- go to DR site VBR and rescan the DR site repository (DR Site VBR now has the latest info on the backup chain in the repo)
- using VLR or FLR functions point the DR site host VEB agent at the DR site VBR
- VEB agent will see the latest (last) backup chain from the main site
- restore away...