Best practice daily with Tape job Monthly

[GianlucaCroci]

Good Morning,
a year ago we implemented an S3 object storage, and we're slowly increasing the Retention to get to have the longest possible period by fully exploiting the storage.
The problem, for which I write here in the Forum, is that at the end of each month we put a FULL of the VMs or Physical Servers on Tape.
These end-of-month jobs to put the saves on Tapes take the saves from the S3 storage that's in the same location as the Tape Library.
To try to reduce this procedure as much as possible, we perform a FULL on S3 every week, so as not to have a chain that is too long and significantly lengthen the save on Tape.

1. what are the best practices for this type of use?
2. how can we've a full every month, which we then put on Tape, and the other days of the month of the incrementals?

Thanks in advance for any suggestions.
Regards
Gianluca

[david.domask]

Hi Gianluca,

There aren't too many practical differences in execution for Backup to Tape regardless of where your backups reside. (i.e., block based vs object storage repositories) Ensure there's a stable connection between proxy and the object storage, and the rest will have no discernible difference for your Backup to Tape strategy.

> how can we've a full every month, which we then put on Tape, and the other days of the month of the incrementals?

Two ideal ways to do this:

1. Normal Backup to Tape job and configure a monthly Virtual Full. This would require the primary (source) job targeting the Object Storage repository is Forever Forward Incremental (no periodic fulls)

2. Tape GFS job with Monthly and Daily media pool configured. Configure a desired monthly day, and on that day a full backup will be placed in the Monthly Media Set. The Daily will have incremental backups, but periodically will require a Full backup be stored in the Daily Media Set. Retention mode for the source job does not matter with Tape GFS

I would not worry about the source chain too much, Backup to Tape (and Tape GFS) are fairly smart in how it handles the source backups to copy to tape and will ensure you can meet your tape-out goals. I'm guessing you maybe remember older versions of the tape integration which were much more reliant on how the source job(s) were configured, but many of those old requirements are no longer essential, Veeam Tape jobs will be able to match your tape-out + backup strategy.

[GianlucaCroci]

Hi David,

thanks a lot.

I was just looking at the second solution, and I think it is the most functional and practical to use.
I've already created a CopyJob by duplicating one, and implementing the Monthly in the GFS instead of the Weekly that I had before. I set it as LAST, which should be what I need.

So if I understood correctly, both from what you wrote to me, but also from what is written on the Veeam site (your link), I will have a FULL with a large chain of incrementals until the next FULL generated by the GFS Monthly. The old chain will then be deleted when the new chain has enough saves corresponding to the value of the Retention.
This will allow me to have less FULL and improve the management of the S3.

correct?
Thanks again

[david.domask]

Hi Gianluca,

You're very welcome for the help.

I think a slight misunderstanding on the recommendation -- Tape GFS will place full backups on tape for each media set on the scheduled day regardless of how the source backup job is configured.

Similarly, remember with Object Storage that Veeam isn't storing normal VBKs + VIBs as standalone files but instead as chunked out data blocks. A full backup may still be useful if your policy requires it, but the idea of a backup chain is largely logical with Object Storage repositories, there is not a long backup chain with VBKs and VIBs.

So for your request, you can do it either by scheduling a periodic full on the source job or leave the source job as Forever Forward Incremental, for the Tape GFS Job it will not matter.

[GianlucaCroci]

Hello,
I've used "Forward Incremental" in the past, but we've about 700 servers that are backed up every day, and about 30 servers have several TB of data to back up, so it would be impractical (many writes to create the new full and the "old" incremental).
I also prefer to take the backups from CopyJob because they are in the same location as the Tape Library, even if I currently use the backups directly because even though it is far away it is still faster.


On this very interesting page
https://helpcenter.veeam.com/archive/ba ... riods.html
I don't quite understand the "Monthly GFS cycle" part.
From what I read, if I choose "Last," in a month with 31 days, the FULL is executed on the 25th? Why?


thanks again

[david.domask]

Understood, Forever Forward then sounds like what you're wanting with Virtual Full backups configured on the tape job as you desire.

[GianlucaCroci]

Or should I select "Monthly" with "First", because that should create a full account on the first day of the month? Correct?

Thanks again

[david.domask]

Correct, if you set it on the 1st, but just noticed you linked the Backup Copy GFS page, and these changes need to be made on the Tape GFS job, not the source job. Tape GFS ignores the source job GFS because Tape GFS has its own schedule and creates the GFS points regardless of the source job configuration.