Copying existing data currently stored on tapes to object storage

[massimiliano.rizzi]

Hello Community and good day,

I am seeking technical advice in order to help an existing customer that is looking at replacing their old Tape Library with an immutable Object Storage based repository in the cloud.

Historically our customer has been using:

==================================================
1. Three backup to tape jobs in order to archive to tape Veeam backups that were produced on disk by Veeam backup jobs
2. A file to tape job in order to back up to tape a bunch of Microsoft Windows files
==================================================

From my perspective, copying new data from now on to an immutable Object Storage based repository in the cloud is pretty straightforward as there's plenty of flexibility especially with the new V12, however I am seeking technical advice with regards to a possible workflow of copying existing data that is currently stored on tapes (at least partially depending on the customer's needs) in order for our customer to be able to decommission the old Tape Library once the existing data that is currently stored on tapes has been offloaded to object storage.

Any advice on this matter will be greatly appreciated.

Kind regards,

Massimiliano

[Mildur]

Hi Massimiliano

You could restore the backups from tape to a disk repository. And then use our Copy backup feature to copy those backups to object storage with the necessary retention period.

Best,
Fabian

[massimiliano.rizzi]

You could restore the backups from tape to a disk repository. And then use our Copy backup feature to copy those backups to object storage with the necessary retention period.

Hi there Fabian,

thank you very much for you reply.

I would like to dive a little deeper into the first step where we actually restore/stage backups stored on Tapes to a local disk repository.

Let's consider a scenario where backups stored on Tapes are created on a monthly basis and usually each of them spans a couple of tapes. Let's say each monthly backup is 1 TB. My understanding is that we will need 1 TB on disk (temporarily if we choose the Move Backup option) for each monthly backup staged on disk and that each staged backup will be visible under Home -> Inventory -> Imported.

If we choose the Move Backup option selecting as a destination a newly created object storage repo, does it make any difference for Veeam whether we stage on disk each monthly backup one at a time and then move it to the newly created object storage repo accordingly ? I mean if, for example, we opt for staging to a local repo all backups stored on Tapes for the past year (January to December 2022), my understanding is that we will (temporarily) need 12 TB on disk for all monthly backups staged on disk and that all staged backups will be visible under Home -> Inventory -> Imported. As all backup files staged on disk belong to the same job, from available upload bandwidth perspective are we going to need to send each staged backup as a complete and independent full or are we going to benefit from source-side deduplication so that when moving backups to the newly created object storage repo Veeam will offload only blocks that are unique to a given restore point, while the rest of blocks will be referenced to the existing blocks already in object storage ? In other words, are we going to need to send 12 TB without benefiting from source-side deduplication ?

Thanks and Regards,

Massimiliano

[Mildur]

Hi Massimiliano

Each restore point restored to a local repository will be listed as it's own backup chain (when Backup to Tape jobs are used).
There won't be any FastClone awareness between different restore points.

Unfortunately we cannot restore the backups from tape and build a new backup chain from it which can be moved with Fast Clone awareness.

Best,
Fabian

[massimiliano.rizzi]

Unfortunately we cannot restore the backups from tape and build a new backup chain from it which can be moved with Fast Clone awareness.

Hi Fabian,

thank you again for you reply.

Each restore point restored to a local repository listed as its own backup chain and, as a result, the inability to copy those backups to object storage with Fast Clone awareness is exactly what I was afraid of, especially from an available upload bandwidth standpoint (rather than from a duplicated space inside the bucket).

As this will not be the case for only unique blocks to be offloaded, costs for API Calls might come into play as well depending on whether the chosen provider bills our customer for both API calls and storage.

As copying existing data that is currently stored on tapes to object storage will be a one-time action, I can see a potential use case for a one-time transfer using an AWS Snowball Edge, Wasabi Ball or similar physical transfer Appliance (depending on the chosen provider and the availability of the transfer Appliance) as the initial offload will be much faster than transferring the same amount of data directly to object storage.

What do you think about that ?

Thanks and Regards,

Massimiliano